winit-sonoma-fix/src/platform_impl/linux/x11/window.rs

1335 lines
49 KiB
Rust
Raw Normal View History

use raw_window_handle::unix::XlibHandle;
use std::{
cmp,
collections::HashSet,
env,
ffi::CString,
mem::{self, replace, MaybeUninit},
os::raw::*,
path::Path,
ptr, slice,
sync::Arc,
};
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
2018-05-08 07:36:21 +10:00
use libc;
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
use parking_lot::Mutex;
use crate::{
dpi::{LogicalPosition, LogicalSize},
error::{ExternalError, NotSupportedError, OsError as RootOsError},
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
monitor::{MonitorHandle as RootMonitorHandle, VideoMode as RootVideoMode},
platform_impl::{
x11::{ime::ImeContextCreationError, MonitorHandle as X11MonitorHandle},
MonitorHandle as PlatformMonitorHandle, OsError, PlatformSpecificWindowBuilderAttributes,
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
VideoMode as PlatformVideoMode,
},
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
window::{CursorIcon, Fullscreen, Icon, WindowAttributes},
};
Move fullscreen modes to not touch physical resolutions (#270) * Fix X11 screen resolution change using XrandR The previous XF86 resolution switching was broken and everything seems to have moved on to xrandr. Use that instead while cleaning up the code a bit as well. * Use XRandR for actual multiscreen support in X11 * Use actual monitor names in X11 * Get rid of ptr::read usage in X11 * Use a bog standard Vec instead of VecDeque * Get rid of the XRandR mode switching stuff Wayland has made the decision that apps shouldn't change screen resolutions and just take the screens as they've been setup. In the modern world where GPU scaling is cheap and LCD panels are scaling anyway it makes no sense to make "physical" resolution changes when software should be taking care of it. This massively simplifies the code and makes it easier to extend to more niche setups like MST and videowalls. * Rename fullscreen options to match new semantics * Implement XRandR 1.5 support * Get rid of the FullScreen enum Moving to just having two states None and Some(MonitorId) and then being able to set full screen in the current monitor with something like: window.set_fullscreen(Some(window.current_monitor())); * Implement Window::get_current_monitor() Do it by iterating over the available monitors and finding which has the biggest overlap with the window. For this MonitorId needs a new get_position() that needs to be implemented for all platforms. * Add unimplemented get_position() to all MonitorId * Make get_current_monitor() platform specific * Add unimplemented get_current_monitor() to all * Implement proper primary monitor selection in X11 * Shut up some warnings * Remove libxxf86vm package from travis Since we're no longer using XF86 there's no need to keep the package around for CI. * Don't use new struct syntax * Fix indentation * Adjust Android/iOS fullscreen/maximized On Android and iOS we can assume single screen apps that are already fullscreen and maximized so there are a few methods that are implemented by just returning a fixed value or not doing anything. * Mark OSX/Win fullscreen/maximized unimplemented()! These would be safe as no-ops but we should make it explicit so there is more of an incentive to actually implement them.
2017-09-07 18:33:46 +10:00
use super::{ffi, util, EventLoopWindowTarget, ImeSender, WindowId, XConnection, XError};
#[derive(Debug)]
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
pub struct SharedState {
2018-05-29 21:48:47 +10:00
pub cursor_pos: Option<(f64, f64)>,
pub size: Option<(u32, u32)>,
pub position: Option<(i32, i32)>,
pub inner_position: Option<(i32, i32)>,
pub inner_position_rel_parent: Option<(i32, i32)>,
pub last_monitor: X11MonitorHandle,
2018-05-29 21:48:47 +10:00
pub dpi_adjusted: Option<(f64, f64)>,
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
pub fullscreen: Option<Fullscreen>,
// Set when application calls `set_fullscreen` when window is not visible
pub desired_fullscreen: Option<Option<Fullscreen>>,
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
// Used to restore position after exiting fullscreen
2018-06-15 09:42:18 +10:00
pub restore_position: Option<(i32, i32)>,
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
// Used to restore video mode after exiting fullscreen
pub desktop_video_mode: Option<(ffi::RRCrtc, ffi::RRMode)>,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
pub frame_extents: Option<util::FrameExtentsHeuristic>,
pub min_inner_size: Option<LogicalSize>,
pub max_inner_size: Option<LogicalSize>,
pub visibility: Visibility,
}
#[derive(Copy, Clone, Debug, Eq, PartialEq)]
pub enum Visibility {
No,
Yes,
// Waiting for VisibilityNotify
YesWait,
2018-06-15 09:42:18 +10:00
}
impl SharedState {
fn new(last_monitor: X11MonitorHandle, is_visible: bool) -> Mutex<Self> {
let visibility = if is_visible {
Visibility::YesWait
} else {
Visibility::No
};
Mutex::new(SharedState {
last_monitor,
visibility,
cursor_pos: None,
size: None,
position: None,
inner_position: None,
inner_position_rel_parent: None,
dpi_adjusted: None,
fullscreen: None,
desired_fullscreen: None,
restore_position: None,
desktop_video_mode: None,
frame_extents: None,
min_inner_size: None,
max_inner_size: None,
})
2018-06-15 09:42:18 +10:00
}
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
}
2018-05-29 21:48:47 +10:00
unsafe impl Send for UnownedWindow {}
unsafe impl Sync for UnownedWindow {}
pub struct UnownedWindow {
pub xconn: Arc<XConnection>, // never changes
xwindow: ffi::Window, // never changes
root: ffi::Window, // never changes
screen_id: i32, // never changes
cursor: Mutex<CursorIcon>,
cursor_grabbed: Mutex<bool>,
cursor_visible: Mutex<bool>,
2018-06-15 09:42:18 +10:00
ime_sender: Mutex<ImeSender>,
2018-05-29 21:48:47 +10:00
pub shared_state: Mutex<SharedState>,
pending_redraws: Arc<::std::sync::Mutex<HashSet<WindowId>>>,
}
2018-05-29 21:48:47 +10:00
impl UnownedWindow {
pub fn new<T>(
event_loop: &EventLoopWindowTarget<T>,
2018-05-08 07:36:21 +10:00
window_attrs: WindowAttributes,
pl_attribs: PlatformSpecificWindowBuilderAttributes,
) -> Result<UnownedWindow, RootOsError> {
2018-05-29 21:48:47 +10:00
let xconn = &event_loop.xconn;
let root = event_loop.root;
let mut monitors = xconn.available_monitors();
let guessed_monitor = if monitors.is_empty() {
X11MonitorHandle::dummy()
} else {
xconn
.query_pointer(root, util::VIRTUAL_CORE_POINTER)
.ok()
.and_then(|pointer_state| {
let (x, y) = (pointer_state.root_x as i64, pointer_state.root_y as i64);
for i in 0..monitors.len() {
if monitors[i].rect.contains_point(x, y) {
return Some(monitors.swap_remove(i));
}
}
None
})
.unwrap_or_else(|| monitors.swap_remove(0))
};
let dpi_factor = guessed_monitor.hidpi_factor();
info!("Guessed window DPI factor: {}", dpi_factor);
let max_inner_size: Option<(u32, u32)> = window_attrs
.max_inner_size
.map(|size| size.to_physical(dpi_factor).into());
let min_inner_size: Option<(u32, u32)> = window_attrs
.min_inner_size
.map(|size| size.to_physical(dpi_factor).into());
2018-06-15 09:42:18 +10:00
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
let dimensions = {
// x11 only applies constraints when the window is actively resized
// by the user, so we have to manually apply the initial constraints
let mut dimensions: (u32, u32) = window_attrs
.inner_size
.or_else(|| Some((800, 600).into()))
.map(|size| size.to_physical(dpi_factor))
2018-06-15 09:42:18 +10:00
.map(Into::into)
.unwrap();
if let Some(max) = max_inner_size {
dimensions.0 = cmp::min(dimensions.0, max.0);
dimensions.1 = cmp::min(dimensions.1, max.1);
}
if let Some(min) = min_inner_size {
dimensions.0 = cmp::max(dimensions.0, min.0);
dimensions.1 = cmp::max(dimensions.1, min.1);
}
debug!(
"Calculated physical dimensions: {}x{}",
dimensions.0, dimensions.1
);
dimensions
};
let screen_id = match pl_attribs.screen_id {
Some(id) => id,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
None => unsafe { (xconn.xlib.XDefaultScreen)(xconn.display) },
};
// creating
let mut set_win_attr = {
let mut swa: ffi::XSetWindowAttributes = unsafe { mem::zeroed() };
swa.colormap = if let Some(vi) = pl_attribs.visual_infos {
unsafe {
let visual = vi.visual;
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
(xconn.xlib.XCreateColormap)(xconn.display, root, visual, ffi::AllocNone)
}
} else {
0
};
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
swa.event_mask = ffi::ExposureMask
| ffi::StructureNotifyMask
| ffi::VisibilityChangeMask
| ffi::KeyPressMask
| ffi::KeyReleaseMask
| ffi::KeymapStateMask
| ffi::ButtonPressMask
| ffi::ButtonReleaseMask
| ffi::PointerMotionMask;
swa.border_pixel = 0;
swa.override_redirect = pl_attribs.override_redirect as c_int;
swa
};
let mut window_attributes = ffi::CWBorderPixel | ffi::CWColormap | ffi::CWEventMask;
2015-06-15 07:20:32 +10:00
if pl_attribs.override_redirect {
window_attributes |= ffi::CWOverrideRedirect;
}
// finally creating the window
2018-05-29 21:48:47 +10:00
let xwindow = unsafe {
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
(xconn.xlib.XCreateWindow)(
xconn.display,
root,
0,
0,
dimensions.0 as c_uint,
dimensions.1 as c_uint,
0,
match pl_attribs.visual_infos {
Some(vi) => vi.depth,
None => ffi::CopyFromParent,
},
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
ffi::InputOutput as c_uint,
Squashed commit of the following: (#853) commit fa95f204d3c10ceca70e794870657a0f33349761 Author: Hal Gentz <zegentzy@protonmail.com> Date: Sun Apr 28 00:14:01 2019 -0600 xrender Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit b62cee51c7b22f6f150bfe04f9b28f024e641323 Merge: 3f021ea7 a6551f46 Author: Hal Gentz <zegentzy@protonmail.com> Date: Thu Apr 25 18:13:43 2019 -0600 Merge branch 'macos-gentz' of github.com:ZeGentzy/winit into macos-gentz commit 3f021ea7f7ac6bc2a697a5b6e4e6424e838a2139 Author: Hal Gentz <zegentzy@protonmail.com> Date: Thu Apr 25 18:04:02 2019 -0600 Get rid of warnings. Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit a6551f4607ea0bc26df8716dee8115371ef367db Author: Hal Gentz <zegentzy@protonmail.com> Date: Thu Apr 25 07:40:56 2019 -0600 Fix example Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit cbfda6c57e9740b49d2b496bda43197f611cb48c Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:47:46 2019 -0600 Fixes Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 86bc86f3d3add4a6125aa9b2eca79061c0dfcd91 Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:39:19 2019 -0600 Backport https://github.com/rust-windowing/winit/commit/9a23ec3c373d096e5e9dd0cc0a23f7baf0b64135#diff-1d95fe39cdbaa708c975380a16c314cb Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 742a688efe2f0eeacc2ffbf49b1157c4aaffccbd Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:09:14 2019 -0600 Backports https://github.com/rust-windowing/winit/commit/45a428141329377a1669698e4ca6561834c0dbcc#diff-1d95fe39cdbaa708c975380a16c314cb Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 6c81f2a517d4e2d5ba2ff3eddca030bce972cb2a Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:05:57 2019 -0600 Francesca's macos changes Also backports https://github.com/rust-windowing/winit/commit/bfbcab3a010667e74fa3cc07ac85bd4c85491e65#diff-1d95fe39cdbaa708c975380a16c314cb commit 7c2e1300c26a0634ad505ce72b90eb6dc2fdcac7 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Apr 24 20:58:26 2019 -0600 Squashed commit of the following: commit 5f4aa9f01a719eef98c6d894801c20ee8f96d30f Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 21 17:14:14 2018 -0500 Protect against reentrancy (messily) commit b75073a5b2a8d65ab8806a00ffee390752255c8c Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 21 15:15:27 2018 -0500 Send resize events immediately commit 8e9fc01bd6b404f59488b130413f48e4e5f89b0d Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 21 16:07:43 2018 -0500 Don't use struct for window delegate commit c6853b0c4a8fe357f463604bb879dc1be424860e Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 21:17:48 2018 -0500 Split up util commit 262c46b148413130fa239099f1151c1f1bd5c13c Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 20:55:00 2018 -0500 Use dispatch crate commit 63152c2f475794d1a36a5b3687c777664d7d5613 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 20:29:13 2018 -0500 RedrawRequested commit 27e475c7c78b059fd9b5e8350cd26756eecdfc94 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 19:24:44 2018 -0500 User events commit 157418d7dedace9c571e977d98ea92464c3188b2 Author: Francesca Plebani <franplebani@gmail.com> Date: Tue Dec 18 22:38:05 2018 -0500 Moved out cursor loading commit b4925641c973979a38743202b4269efe09ac43b4 Author: Francesca Plebani <franplebani@gmail.com> Date: Tue Dec 18 21:32:12 2018 -0500 Fixed a bunch of threading issues commit 4aef63dfb78dfaf38c83cb0e88d4ea9d8d0578a6 Author: Francesca Plebani <franplebani@gmail.com> Date: Mon Dec 17 13:54:59 2018 -0500 Wait works commit 72ed426c695df5dc410902263bd74188059b8ddd Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 14 20:49:10 2018 -0500 Fixed drag and dropg commit 658209f4a20acd536218f41a01fb8cbbebc705e0 Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 14 20:42:42 2018 -0500 Made mutexes finer for less deadlock risk commit 8e6b9866084690da900c4d058e412cab8ebb30c4 Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 14 16:45:06 2018 -0500 Dump (encapsulate) everything into AppState commit d2dc83df15939d89301e2cff0ffa2d98c48b406f Author: Francesca Plebani <franplebani@gmail.com> Date: Thu Dec 13 17:36:47 2018 -0500 All window events work! commit 7c7fcc98872b3c35bd7767b5c6235a74bc105e06 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 12 17:11:09 2018 -0500 Very rough usage of CFRunLoop commit 3c7a52ff4df683b5b7e1751e4051ec445a818774 Author: Francesca Plebani <franplebani@gmail.com> Date: Tue Dec 11 15:45:23 2018 -0500 Fixed deadlocks commit b74c7fe1bcd173e9b0c0e004956c257e805bc2a2 Author: Francesca Plebani <franplebani@gmail.com> Date: Mon Dec 10 18:59:46 2018 -0500 Fix keyDown deadlock commit 3798f9c1a4bef2a3d1552f846b26efc31b1bbb6c Author: Francesca Plebani <franplebani@gmail.com> Date: Mon Dec 10 18:44:40 2018 -0500 It builds! commit 8c8620214357714c5cd0b3beefda6704512e3f64 Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 7 21:09:55 2018 -0500 Horribly broken so far commit 8269ed2a9270e5ec5b14f80fd21d1e0e6f51be29 Author: Osspial <osspial@gmail.com> Date: Mon Nov 19 23:51:20 2018 -0500 Fix crash with runner refcell not getting dropped commit 54ce6a21a0722e408ae49c74f5008005fc1e4cbf Author: Osspial <osspial@gmail.com> Date: Sun Nov 18 19:12:45 2018 -0500 Fix buffered events not getting dispatched commit 2c18b804df66f49f93cfe722a679d6c5e01d8cb1 Author: Osspial <osspial@gmail.com> Date: Sun Nov 18 18:51:24 2018 -0500 Fix thread executor not executing closure when called from non-loop thread commit 5a3a5e2293cec3e566c4aac344ae7eaa343608b5 Author: Osspial <osspial@gmail.com> Date: Thu Nov 15 22:43:59 2018 -0500 Fix some deadlocks that could occur when changing window state commit 2a3cefd8c5df1c06127b05651cbdf5e3d9e3a6d3 Author: Osspial <osspial@gmail.com> Date: Thu Nov 15 16:45:17 2018 -0500 Document and implement Debug for EventLoopWindowTarget commit fa46825a289ca0587dc97f9c00dea5516fb4925a Author: Osspial <osspial@gmail.com> Date: Thu Nov 15 16:40:48 2018 -0500 Replace &EventLoop in callback with &EventLoopWindowTarget commit 9f36a7a68e1dc379cf9091213dae2c3586d3e473 Author: Osspial <osspial@gmail.com> Date: Wed Nov 14 21:28:38 2018 -0500 Fix freeze when setting decorations commit d9c3daca9b459e02ef614568fe803a723965fe8d Author: Osspial <osspial@gmail.com> Date: Fri Nov 9 20:41:15 2018 -0500 Fix 1.24.1 build commit 5289d22372046bac403a279c3641737c0cfc46d2 Author: Osspial <osspial@gmail.com> Date: Fri Nov 9 00:00:27 2018 -0500 Remove serde implementations from ControlFlow commit 92ac3d6ac7915923c22c380cc3a74c5f3830708e Author: Osspial <osspial@gmail.com> Date: Thu Nov 8 23:46:41 2018 -0500 Remove crossbeam dependency and make drop events work again commit 8299eb2f03773a34079c61fc8adb51405aafc467 Author: Osspial <osspial@gmail.com> Date: Thu Sep 13 22:39:40 2018 -0400 Fix crash when running in release mode commit bb6ab1bb6e9595e90f1915fdde7e23904f2ba594 Author: Osspial <osspial@gmail.com> Date: Sun Sep 9 14:28:16 2018 -0400 Fix unreachable panic after setting ControlFlow to Poll during some RedrawRequested events. commit 5068ff4ee152bfe93c9190235f02d001202feb88 Author: Osspial <osspial@gmail.com> Date: Sun Sep 9 14:14:28 2018 -0400 Improve clarity/fix typos in docs commit 8ed575ff4a4f0961bb2e784bda1ae109c6bd37b7 Author: Osspial <osspial@gmail.com> Date: Sun Sep 9 00:19:53 2018 -0400 Update send test and errors that broke some examples/APIs commit bf7bfa82ebb5d6ae110ce0492c124ef462945f85 Author: Osspial <osspial@gmail.com> Date: Wed Sep 5 22:36:05 2018 -0400 Fix resize lag when waiting in some situations commit 70722cc4c322e3e599b3a03bce5058a5d433970b Author: Osspial <osspial@gmail.com> Date: Wed Sep 5 19:58:52 2018 -1100 When SendEvent is called during event closure, buffer events commit 53370924b25da15ddd172173150b228065324864 Author: Osspial <osspial@gmail.com> Date: Sun Aug 26 21:55:51 2018 -0400 Improve WaitUntil timer precision commit a654400e730400c2e3584be2f47153043b5b7efe Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 21:06:19 2018 -0400 Add CHANGELOG entry commit deb7d379b7c04e61d6d50ff655eccac0ad692e44 Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 20:19:56 2018 -0400 Rename MonitorId to MonitorHandle commit 8d8d9b7cd1386c99c40023d86e17d10c3fd6652f Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 20:16:52 2018 -0400 Change instances of "events_loop" to "event_loop" commit 0f344087630ae252c9c8f453864e684a1a5405b1 Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 20:13:53 2018 -0400 Improve docs for run and run_return commit fba41f7a7ed8585cbb658b6d0b2f34f75482cb3d Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 19:09:53 2018 -0400 Small changes to examples commit 42e8a0d2cf77af79da082fff7cd29cc8f52d99df Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 19:09:19 2018 -0400 Improve documentation commit 4377680a44ea86dad52954f90bc7d8ad7ed0b4bf Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 23:01:36 2018 -0400 Re-organize into module structure commit f20fac99f6ac57c51603a92d792fd4f665feb7f6 Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 22:07:39 2018 -0400 Add platform::desktop module with EventLoopExt::run_return commit dad24d086aaaff60e557efc4f41d1ae7e3c71738 Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 18:03:41 2018 -0400 Rename os to platform, add Ext trait postfixes commit 7df59c60a06008226f6455619e7242ed0156ed8d Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 17:59:36 2018 -0400 Rename platform to platform_impl commit 99c0f84a9fc771c9c96099232de3716ddf27ca80 Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 17:55:27 2018 -0400 Add request_redraw commit a0fef1a5fad9b5d5da59fff191c7d9c398ea9e01 Author: Osspial <osspial@gmail.com> Date: Mon Aug 20 01:47:11 2018 -0400 Fully invert windows control flow so win32 calls into winit's callback commit 2c607ff87f8fbcad8aa9dc3783b3298c014dd177 Author: Osspial <osspial@gmail.com> Date: Sun Aug 19 13:44:22 2018 -0400 Add ability to send custom user events commit a0b2bb36953f018ff782cef8fc86c6db9343095d Author: Osspial <osspial@gmail.com> Date: Fri Aug 17 17:49:46 2018 -0400 Add StartCause::Init support, timer example commit 02f922f003f56215b92b8feeb9148ad2dd181fc2 Author: Osspial <osspial@gmail.com> Date: Fri Aug 17 17:31:04 2018 -0400 Implement new ControlFlow and associated events commit 8b8a7675ec67e15a0f8f69db0bdeb79bee0ac20d Author: Osspial <osspial@gmail.com> Date: Fri Jul 13 01:48:26 2018 -0400 Replace windows Mutex with parking_lot Mutex commit 9feada206f6b9fb1e9da118be6b77dfc217ace8d Author: Osspial <osspial@gmail.com> Date: Fri Jul 13 01:39:53 2018 -0400 Update run_forever to hijack thread commit 2e83bac99cc264cd2723cb182feea84a0a15e08d Author: Osspial <osspial@gmail.com> Date: Thu Jul 12 23:43:58 2018 -0400 Remove second thread from win32 backend commit 64b8a9c6a50362d10c074077a1e37b057f3e3c81 Author: Osspial <osspial@gmail.com> Date: Thu Jul 12 22:13:07 2018 -0400 Rename WindowEvent::Refresh to WindowEvent::Redraw commit 529c08555fd0b709a23d486211d28fbd0980fc94 Author: Osspial <osspial@gmail.com> Date: Thu Jul 12 22:04:38 2018 -0400 Rename EventsLoop and associated types to EventLoop Signed-off-by: Hal Gentz <zegentzy@protonmail.com> Co-authored-by: Hal Gentz <zegentzy@protonmail.com> commit cfb929ba0a9e787f8bb1a6dae4e05e4c7776bc97 Author: Hal Gentz <zegentzy@protonmail.com> Date: Thu Apr 25 07:40:56 2019 -0600 Fix example Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 68d3317ff58381d55f5f9bd3db0860d66544fe12 Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:47:46 2019 -0600 Fixes Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 02d1aae4db27df054b703aa935ca118f31e17123 Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:39:19 2019 -0600 Backport https://github.com/rust-windowing/winit/commit/9a23ec3c373d096e5e9dd0cc0a23f7baf0b64135#diff-1d95fe39cdbaa708c975380a16c314cb Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit dd9de5a6d444a9ab17afe470f4cf2a57e3ed76ae Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:09:14 2019 -0600 Backports https://github.com/rust-windowing/winit/commit/45a428141329377a1669698e4ca6561834c0dbcc#diff-1d95fe39cdbaa708c975380a16c314cb Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 533e2adc1d1e417742475786635848b1620e476c Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 23:05:57 2019 -0600 Francesca's macos changes Also backports https://github.com/rust-windowing/winit/commit/bfbcab3a010667e74fa3cc07ac85bd4c85491e65#diff-1d95fe39cdbaa708c975380a16c314cb commit 73b52221080bd3a881ae3a58c2dbb19bc8d954c6 Author: Hal Gentz <zegentzy@protonmail.com> Date: Wed Apr 24 20:58:26 2019 -0600 Squashed commit of the following: commit 5f4aa9f01a719eef98c6d894801c20ee8f96d30f Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 21 17:14:14 2018 -0500 Protect against reentrancy (messily) commit b75073a5b2a8d65ab8806a00ffee390752255c8c Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 21 15:15:27 2018 -0500 Send resize events immediately commit 8e9fc01bd6b404f59488b130413f48e4e5f89b0d Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 21 16:07:43 2018 -0500 Don't use struct for window delegate commit c6853b0c4a8fe357f463604bb879dc1be424860e Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 21:17:48 2018 -0500 Split up util commit 262c46b148413130fa239099f1151c1f1bd5c13c Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 20:55:00 2018 -0500 Use dispatch crate commit 63152c2f475794d1a36a5b3687c777664d7d5613 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 20:29:13 2018 -0500 RedrawRequested commit 27e475c7c78b059fd9b5e8350cd26756eecdfc94 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 19 19:24:44 2018 -0500 User events commit 157418d7dedace9c571e977d98ea92464c3188b2 Author: Francesca Plebani <franplebani@gmail.com> Date: Tue Dec 18 22:38:05 2018 -0500 Moved out cursor loading commit b4925641c973979a38743202b4269efe09ac43b4 Author: Francesca Plebani <franplebani@gmail.com> Date: Tue Dec 18 21:32:12 2018 -0500 Fixed a bunch of threading issues commit 4aef63dfb78dfaf38c83cb0e88d4ea9d8d0578a6 Author: Francesca Plebani <franplebani@gmail.com> Date: Mon Dec 17 13:54:59 2018 -0500 Wait works commit 72ed426c695df5dc410902263bd74188059b8ddd Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 14 20:49:10 2018 -0500 Fixed drag and dropg commit 658209f4a20acd536218f41a01fb8cbbebc705e0 Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 14 20:42:42 2018 -0500 Made mutexes finer for less deadlock risk commit 8e6b9866084690da900c4d058e412cab8ebb30c4 Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 14 16:45:06 2018 -0500 Dump (encapsulate) everything into AppState commit d2dc83df15939d89301e2cff0ffa2d98c48b406f Author: Francesca Plebani <franplebani@gmail.com> Date: Thu Dec 13 17:36:47 2018 -0500 All window events work! commit 7c7fcc98872b3c35bd7767b5c6235a74bc105e06 Author: Francesca Plebani <franplebani@gmail.com> Date: Wed Dec 12 17:11:09 2018 -0500 Very rough usage of CFRunLoop commit 3c7a52ff4df683b5b7e1751e4051ec445a818774 Author: Francesca Plebani <franplebani@gmail.com> Date: Tue Dec 11 15:45:23 2018 -0500 Fixed deadlocks commit b74c7fe1bcd173e9b0c0e004956c257e805bc2a2 Author: Francesca Plebani <franplebani@gmail.com> Date: Mon Dec 10 18:59:46 2018 -0500 Fix keyDown deadlock commit 3798f9c1a4bef2a3d1552f846b26efc31b1bbb6c Author: Francesca Plebani <franplebani@gmail.com> Date: Mon Dec 10 18:44:40 2018 -0500 It builds! commit 8c8620214357714c5cd0b3beefda6704512e3f64 Author: Francesca Plebani <franplebani@gmail.com> Date: Fri Dec 7 21:09:55 2018 -0500 Horribly broken so far commit 8269ed2a9270e5ec5b14f80fd21d1e0e6f51be29 Author: Osspial <osspial@gmail.com> Date: Mon Nov 19 23:51:20 2018 -0500 Fix crash with runner refcell not getting dropped commit 54ce6a21a0722e408ae49c74f5008005fc1e4cbf Author: Osspial <osspial@gmail.com> Date: Sun Nov 18 19:12:45 2018 -0500 Fix buffered events not getting dispatched commit 2c18b804df66f49f93cfe722a679d6c5e01d8cb1 Author: Osspial <osspial@gmail.com> Date: Sun Nov 18 18:51:24 2018 -0500 Fix thread executor not executing closure when called from non-loop thread commit 5a3a5e2293cec3e566c4aac344ae7eaa343608b5 Author: Osspial <osspial@gmail.com> Date: Thu Nov 15 22:43:59 2018 -0500 Fix some deadlocks that could occur when changing window state commit 2a3cefd8c5df1c06127b05651cbdf5e3d9e3a6d3 Author: Osspial <osspial@gmail.com> Date: Thu Nov 15 16:45:17 2018 -0500 Document and implement Debug for EventLoopWindowTarget commit fa46825a289ca0587dc97f9c00dea5516fb4925a Author: Osspial <osspial@gmail.com> Date: Thu Nov 15 16:40:48 2018 -0500 Replace &EventLoop in callback with &EventLoopWindowTarget commit 9f36a7a68e1dc379cf9091213dae2c3586d3e473 Author: Osspial <osspial@gmail.com> Date: Wed Nov 14 21:28:38 2018 -0500 Fix freeze when setting decorations commit d9c3daca9b459e02ef614568fe803a723965fe8d Author: Osspial <osspial@gmail.com> Date: Fri Nov 9 20:41:15 2018 -0500 Fix 1.24.1 build commit 5289d22372046bac403a279c3641737c0cfc46d2 Author: Osspial <osspial@gmail.com> Date: Fri Nov 9 00:00:27 2018 -0500 Remove serde implementations from ControlFlow commit 92ac3d6ac7915923c22c380cc3a74c5f3830708e Author: Osspial <osspial@gmail.com> Date: Thu Nov 8 23:46:41 2018 -0500 Remove crossbeam dependency and make drop events work again commit 8299eb2f03773a34079c61fc8adb51405aafc467 Author: Osspial <osspial@gmail.com> Date: Thu Sep 13 22:39:40 2018 -0400 Fix crash when running in release mode commit bb6ab1bb6e9595e90f1915fdde7e23904f2ba594 Author: Osspial <osspial@gmail.com> Date: Sun Sep 9 14:28:16 2018 -0400 Fix unreachable panic after setting ControlFlow to Poll during some RedrawRequested events. commit 5068ff4ee152bfe93c9190235f02d001202feb88 Author: Osspial <osspial@gmail.com> Date: Sun Sep 9 14:14:28 2018 -0400 Improve clarity/fix typos in docs commit 8ed575ff4a4f0961bb2e784bda1ae109c6bd37b7 Author: Osspial <osspial@gmail.com> Date: Sun Sep 9 00:19:53 2018 -0400 Update send test and errors that broke some examples/APIs commit bf7bfa82ebb5d6ae110ce0492c124ef462945f85 Author: Osspial <osspial@gmail.com> Date: Wed Sep 5 22:36:05 2018 -0400 Fix resize lag when waiting in some situations commit 70722cc4c322e3e599b3a03bce5058a5d433970b Author: Osspial <osspial@gmail.com> Date: Wed Sep 5 19:58:52 2018 -1100 When SendEvent is called during event closure, buffer events commit 53370924b25da15ddd172173150b228065324864 Author: Osspial <osspial@gmail.com> Date: Sun Aug 26 21:55:51 2018 -0400 Improve WaitUntil timer precision commit a654400e730400c2e3584be2f47153043b5b7efe Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 21:06:19 2018 -0400 Add CHANGELOG entry commit deb7d379b7c04e61d6d50ff655eccac0ad692e44 Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 20:19:56 2018 -0400 Rename MonitorId to MonitorHandle commit 8d8d9b7cd1386c99c40023d86e17d10c3fd6652f Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 20:16:52 2018 -0400 Change instances of "events_loop" to "event_loop" commit 0f344087630ae252c9c8f453864e684a1a5405b1 Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 20:13:53 2018 -0400 Improve docs for run and run_return commit fba41f7a7ed8585cbb658b6d0b2f34f75482cb3d Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 19:09:53 2018 -0400 Small changes to examples commit 42e8a0d2cf77af79da082fff7cd29cc8f52d99df Author: Osspial <osspial@gmail.com> Date: Thu Aug 23 19:09:19 2018 -0400 Improve documentation commit 4377680a44ea86dad52954f90bc7d8ad7ed0b4bf Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 23:01:36 2018 -0400 Re-organize into module structure commit f20fac99f6ac57c51603a92d792fd4f665feb7f6 Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 22:07:39 2018 -0400 Add platform::desktop module with EventLoopExt::run_return commit dad24d086aaaff60e557efc4f41d1ae7e3c71738 Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 18:03:41 2018 -0400 Rename os to platform, add Ext trait postfixes commit 7df59c60a06008226f6455619e7242ed0156ed8d Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 17:59:36 2018 -0400 Rename platform to platform_impl commit 99c0f84a9fc771c9c96099232de3716ddf27ca80 Author: Osspial <osspial@gmail.com> Date: Wed Aug 22 17:55:27 2018 -0400 Add request_redraw commit a0fef1a5fad9b5d5da59fff191c7d9c398ea9e01 Author: Osspial <osspial@gmail.com> Date: Mon Aug 20 01:47:11 2018 -0400 Fully invert windows control flow so win32 calls into winit's callback commit 2c607ff87f8fbcad8aa9dc3783b3298c014dd177 Author: Osspial <osspial@gmail.com> Date: Sun Aug 19 13:44:22 2018 -0400 Add ability to send custom user events commit a0b2bb36953f018ff782cef8fc86c6db9343095d Author: Osspial <osspial@gmail.com> Date: Fri Aug 17 17:49:46 2018 -0400 Add StartCause::Init support, timer example commit 02f922f003f56215b92b8feeb9148ad2dd181fc2 Author: Osspial <osspial@gmail.com> Date: Fri Aug 17 17:31:04 2018 -0400 Implement new ControlFlow and associated events commit 8b8a7675ec67e15a0f8f69db0bdeb79bee0ac20d Author: Osspial <osspial@gmail.com> Date: Fri Jul 13 01:48:26 2018 -0400 Replace windows Mutex with parking_lot Mutex commit 9feada206f6b9fb1e9da118be6b77dfc217ace8d Author: Osspial <osspial@gmail.com> Date: Fri Jul 13 01:39:53 2018 -0400 Update run_forever to hijack thread commit 2e83bac99cc264cd2723cb182feea84a0a15e08d Author: Osspial <osspial@gmail.com> Date: Thu Jul 12 23:43:58 2018 -0400 Remove second thread from win32 backend commit 64b8a9c6a50362d10c074077a1e37b057f3e3c81 Author: Osspial <osspial@gmail.com> Date: Thu Jul 12 22:13:07 2018 -0400 Rename WindowEvent::Refresh to WindowEvent::Redraw commit 529c08555fd0b709a23d486211d28fbd0980fc94 Author: Osspial <osspial@gmail.com> Date: Thu Jul 12 22:04:38 2018 -0400 Rename EventsLoop and associated types to EventLoop Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit ab1dfaaaa53a3acd206bf494ac90e3fe130dc609 Author: Hal Gentz <zegentzy@protonmail.com> Date: Tue Apr 23 21:52:17 2019 -0600 Minor Signed-off-by: Hal Gentz <zegentzy@protonmail.com> commit 7933209d603e0794adb806d9cf53507f1c2f1d3c Author: Victor Berger <victor.berger@m4x.org> Date: Thu Apr 18 09:10:41 2019 +0200 wayland/x11: Make ControlFlow::Exit sticky commit 8355a7513e299ffba21062c8518bcf4bdb735ba9 Author: Victor Berger <victor.berger@m4x.org> Date: Tue Apr 16 12:21:33 2019 +0200 x11: Implement run_return using calloop commit f64edb60cc85fcd98a1cec955ba9980f617fdd73 Author: Victor Berger <victor.berger@m4x.org> Date: Tue Apr 16 10:42:04 2019 +0200 x11: port to evl2 with stubs commit be372898ddc60e47887c9a152c10ff498445f8cf Author: Victor Berger <victor.berger@m4x.org> Date: Mon Apr 15 17:35:59 2019 +0200 Fix compilation on Linux. Signed-off-by: Hal Gentz <zegentzy@protonmail.com> Co-authored-by: Francesca Plebani <franplebani@gmail.com>
2019-05-02 09:03:30 +10:00
// TODO: If window wants transparency and `visual_infos` is None,
// we need to find our own visual which has an `alphaMask` which
// is > 0, like we do in glutin.
//
// It is non obvious which masks, if any, we should pass to
// `XGetVisualInfo`. winit doesn't recieve any info about what
// properties the user wants. Users should consider choosing the
// visual themselves as glutin does.
match pl_attribs.visual_infos {
Some(vi) => vi.visual,
None => ffi::CopyFromParent as *mut ffi::Visual,
},
window_attributes,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
&mut set_win_attr,
)
};
2018-05-29 21:48:47 +10:00
let window = UnownedWindow {
xconn: Arc::clone(xconn),
xwindow,
root,
screen_id,
2018-05-29 21:48:47 +10:00
cursor: Default::default(),
cursor_grabbed: Mutex::new(false),
cursor_visible: Mutex::new(true),
2018-06-15 09:42:18 +10:00
ime_sender: Mutex::new(event_loop.ime_sender.clone()),
shared_state: SharedState::new(guessed_monitor, window_attrs.visible),
pending_redraws: event_loop.pending_redraws.clone(),
};
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
// Title must be set before mapping. Some tiling window managers (i.e. i3) use the window
// title to determine placement/etc., so doing this after mapping would cause the WM to
// act on the wrong title state.
window.set_title_inner(&window_attrs.title).queue();
window
.set_decorations_inner(window_attrs.decorations)
.queue();
{
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
// Enable drag and drop (TODO: extend API to make this toggleable)
unsafe {
2018-05-27 22:49:35 +10:00
let dnd_aware_atom = xconn.get_atom_unchecked(b"XdndAware\0");
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
let version = &[5 as c_ulong]; // Latest version; hasn't changed since 2002
2018-05-27 22:49:35 +10:00
xconn.change_property(
2018-05-29 21:48:47 +10:00
window.xwindow,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
dnd_aware_atom,
ffi::XA_ATOM,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
util::PropMode::Replace,
version,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
)
}
.queue();
2015-12-24 20:57:08 +11:00
// WM_CLASS must be set *before* mapping the window, as per ICCCM!
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
{
let (class, instance) = if let Some((instance, class)) = pl_attribs.class {
let instance = CString::new(instance.as_str())
.expect("`WM_CLASS` instance contained null byte");
let class =
CString::new(class.as_str()).expect("`WM_CLASS` class contained null byte");
(instance, class)
} else {
let class = env::args()
.next()
.as_ref()
// Default to the name of the binary (via argv[0])
.and_then(|path| Path::new(path).file_name())
.and_then(|bin_name| bin_name.to_str())
.map(|bin_name| bin_name.to_owned())
.or_else(|| Some(window_attrs.title.clone()))
.and_then(|string| CString::new(string.as_str()).ok())
.expect("Default `WM_CLASS` class contained null byte");
// This environment variable is extraordinarily unlikely to actually be used...
let instance = env::var("RESOURCE_NAME")
.ok()
.and_then(|instance| CString::new(instance.as_str()).ok())
.or_else(|| Some(class.clone()))
.expect("Default `WM_CLASS` instance contained null byte");
(instance, class)
};
2018-05-27 22:49:35 +10:00
let mut class_hint = xconn.alloc_class_hint();
(*class_hint).res_name = class.as_ptr() as *mut c_char;
(*class_hint).res_class = instance.as_ptr() as *mut c_char;
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
unsafe {
(xconn.xlib.XSetClassHint)(xconn.display, window.xwindow, class_hint.ptr);
} //.queue();
}
2018-05-29 21:48:47 +10:00
window.set_pid().map(|flusher| flusher.queue());
window.set_window_types(pl_attribs.x11_window_types).queue();
if let Some(variant) = pl_attribs.gtk_theme_variant {
window.set_gtk_theme_variant(variant).queue();
}
// set size hints
{
let mut min_inner_size = window_attrs
.min_inner_size
.map(|size| size.to_physical(dpi_factor));
let mut max_inner_size = window_attrs
.max_inner_size
.map(|size| size.to_physical(dpi_factor));
if !window_attrs.resizable {
if util::wm_name_is_one_of(&["Xfwm4"]) {
warn!("To avoid a WM bug, disabling resizing has no effect on Xfwm4");
} else {
max_inner_size = Some(dimensions.into());
min_inner_size = Some(dimensions.into());
let mut shared_state_lock = window.shared_state.lock();
shared_state_lock.min_inner_size = window_attrs.min_inner_size;
shared_state_lock.max_inner_size = window_attrs.max_inner_size;
}
}
2018-06-15 09:42:18 +10:00
let mut normal_hints = util::NormalHints::new(xconn);
normal_hints.set_size(Some(dimensions));
normal_hints.set_min_size(min_inner_size.map(Into::into));
normal_hints.set_max_size(max_inner_size.map(Into::into));
2018-06-15 09:42:18 +10:00
normal_hints.set_resize_increments(pl_attribs.resize_increments);
normal_hints.set_base_size(pl_attribs.base_size);
xconn.set_normal_hints(window.xwindow, normal_hints).queue();
}
2018-05-08 07:36:21 +10:00
// Set window icons
if let Some(icon) = window_attrs.window_icon {
window.set_icon_inner(icon).queue();
}
// Opt into handling window close
unsafe {
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
(xconn.xlib.XSetWMProtocols)(
xconn.display,
2018-05-29 21:48:47 +10:00
window.xwindow,
2019-06-28 10:40:27 +10:00
&[event_loop.wm_delete_window, event_loop.net_wm_ping] as *const ffi::Atom
as *mut ffi::Atom,
2,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
);
} //.queue();
// Set visibility (map window)
if window_attrs.visible {
unsafe {
2018-05-29 21:48:47 +10:00
(xconn.xlib.XMapRaised)(xconn.display, window.xwindow);
} //.queue();
}
// Attempt to make keyboard input repeat detectable
unsafe {
let mut supported_ptr = ffi::False;
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
(xconn.xlib.XkbSetDetectableAutoRepeat)(
xconn.display,
ffi::True,
&mut supported_ptr,
);
if supported_ptr == ffi::False {
return Err(os_error!(OsError::XMisc(
"`XkbSetDetectableAutoRepeat` failed"
)));
}
}
// Select XInput2 events
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
let mask = {
let mask = ffi::XI_MotionMask
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
| ffi::XI_ButtonPressMask
| ffi::XI_ButtonReleaseMask
//| ffi::XI_KeyPressMask
//| ffi::XI_KeyReleaseMask
| ffi::XI_EnterMask
| ffi::XI_LeaveMask
| ffi::XI_FocusInMask
| ffi::XI_FocusOutMask
| ffi::XI_TouchBeginMask
| ffi::XI_TouchUpdateMask
| ffi::XI_TouchEndMask;
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
mask
};
xconn
.select_xinput_events(window.xwindow, ffi::XIAllMasterDevices, mask)
.queue();
2018-06-15 09:42:18 +10:00
{
let result = event_loop.ime.borrow_mut().create_context(window.xwindow);
2018-06-15 09:42:18 +10:00
if let Err(err) = result {
let e = match err {
ImeContextCreationError::XError(err) => OsError::XError(err),
ImeContextCreationError::Null => {
OsError::XMisc("IME Context creation failed")
2019-06-25 02:14:55 +10:00
}
};
return Err(os_error!(e));
2018-06-15 09:42:18 +10:00
}
}
// These properties must be set after mapping
if window_attrs.maximized {
window.set_maximized_inner(window_attrs.maximized).queue();
}
if window_attrs.fullscreen.is_some() {
window
.set_fullscreen_inner(window_attrs.fullscreen.clone())
.map(|flusher| flusher.queue());
}
if window_attrs.always_on_top {
window
.set_always_on_top_inner(window_attrs.always_on_top)
.queue();
}
}
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
// We never want to give the user a broken window, since by then, it's too late to handle.
xconn
.sync_with_server()
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
.map(|_| window)
.map_err(|x_err| os_error!(OsError::XError(x_err)))
}
2018-06-15 09:42:18 +10:00
fn logicalize_coords(&self, (x, y): (i32, i32)) -> LogicalPosition {
let dpi = self.hidpi_factor();
2018-06-15 09:42:18 +10:00
LogicalPosition::from_physical((x, y), dpi)
}
fn logicalize_size(&self, (width, height): (u32, u32)) -> LogicalSize {
let dpi = self.hidpi_factor();
2018-06-15 09:42:18 +10:00
LogicalSize::from_physical((width, height), dpi)
}
fn set_pid(&self) -> Option<util::Flusher<'_>> {
2018-05-29 21:48:47 +10:00
let pid_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_PID\0") };
let client_machine_atom = unsafe { self.xconn.get_atom_unchecked(b"WM_CLIENT_MACHINE\0") };
unsafe {
// 64 would suffice for Linux, but 256 will be enough everywhere (as per SUSv2). For instance, this is
// the limit defined by OpenBSD.
const MAXHOSTNAMELEN: usize = 256;
// `assume_init` is safe here because the array consists of `MaybeUninit` values,
// which do not require initialization.
let mut buffer: [MaybeUninit<c_char>; MAXHOSTNAMELEN] =
MaybeUninit::uninit().assume_init();
let status = libc::gethostname(buffer.as_mut_ptr() as *mut c_char, buffer.len());
if status != 0 {
return None;
}
ptr::write(buffer[MAXHOSTNAMELEN - 1].as_mut_ptr() as *mut u8, b'\0'); // a little extra safety
let hostname_length = libc::strlen(buffer.as_ptr() as *const c_char);
let hostname = slice::from_raw_parts(buffer.as_ptr() as *const c_char, hostname_length);
self.xconn
.change_property(
self.xwindow,
pid_atom,
ffi::XA_CARDINAL,
util::PropMode::Replace,
&[libc::getpid() as util::Cardinal],
)
.queue();
2018-05-29 21:48:47 +10:00
let flusher = self.xconn.change_property(
self.xwindow,
client_machine_atom,
ffi::XA_STRING,
util::PropMode::Replace,
&hostname[0..hostname_length],
);
Some(flusher)
}
}
fn set_window_types(&self, window_types: Vec<util::WindowType>) -> util::Flusher<'_> {
2018-05-29 21:48:47 +10:00
let hint_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_WINDOW_TYPE\0") };
let atoms: Vec<_> = window_types
.iter()
.map(|t| t.as_atom(&self.xconn))
.collect();
2018-05-29 21:48:47 +10:00
self.xconn.change_property(
self.xwindow,
2018-05-27 22:49:35 +10:00
hint_atom,
ffi::XA_ATOM,
util::PropMode::Replace,
&atoms,
2018-05-27 22:49:35 +10:00
)
}
fn set_gtk_theme_variant(&self, variant: String) -> util::Flusher<'_> {
let hint_atom = unsafe { self.xconn.get_atom_unchecked(b"_GTK_THEME_VARIANT\0") };
let utf8_atom = unsafe { self.xconn.get_atom_unchecked(b"UTF8_STRING\0") };
let variant = CString::new(variant).expect("`_GTK_THEME_VARIANT` contained null byte");
self.xconn.change_property(
self.xwindow,
hint_atom,
utf8_atom,
util::PropMode::Replace,
variant.as_bytes(),
)
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_urgent(&self, is_urgent: bool) {
let mut wm_hints = self
.xconn
.get_wm_hints(self.xwindow)
.expect("`XGetWMHints` failed");
if is_urgent {
(*wm_hints).flags |= ffi::XUrgencyHint;
} else {
(*wm_hints).flags &= !ffi::XUrgencyHint;
}
self.xconn
.set_wm_hints(self.xwindow, wm_hints)
.flush()
.expect("Failed to set urgency hint");
}
fn set_netwm(
2018-05-29 21:48:47 +10:00
&self,
operation: util::StateOperation,
properties: (c_long, c_long, c_long, c_long),
) -> util::Flusher<'_> {
2018-05-29 21:48:47 +10:00
let state_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_STATE\0") };
self.xconn.send_client_msg(
self.xwindow,
self.root,
2018-05-27 22:49:35 +10:00
state_atom,
Some(ffi::SubstructureRedirectMask | ffi::SubstructureNotifyMask),
[
operation as c_long,
properties.0,
properties.1,
properties.2,
properties.3,
],
)
}
fn set_fullscreen_hint(&self, fullscreen: bool) -> util::Flusher<'_> {
let fullscreen_atom =
unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_STATE_FULLSCREEN\0") };
let flusher = self.set_netwm(fullscreen.into(), (fullscreen_atom as c_long, 0, 0, 0));
if fullscreen {
// Ensure that the fullscreen window receives input focus to prevent
// locking up the user's display.
unsafe {
(self.xconn.xlib.XSetInputFocus)(
self.xconn.display,
self.xwindow,
ffi::RevertToParent,
ffi::CurrentTime,
);
}
}
flusher
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
}
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
fn set_fullscreen_inner(&self, fullscreen: Option<Fullscreen>) -> Option<util::Flusher<'_>> {
let mut shared_state_lock = self.shared_state.lock();
match shared_state_lock.visibility {
// Setting fullscreen on a window that is not visible will generate an error.
Visibility::No | Visibility::YesWait => {
shared_state_lock.desired_fullscreen = Some(fullscreen);
return None;
}
Visibility::Yes => (),
}
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
let old_fullscreen = shared_state_lock.fullscreen.clone();
if old_fullscreen == fullscreen {
return None;
}
shared_state_lock.fullscreen = fullscreen.clone();
match (&old_fullscreen, &fullscreen) {
// Store the desktop video mode before entering exclusive
// fullscreen, so we can restore it upon exit, as XRandR does not
// provide a mechanism to set this per app-session or restore this
// to the desktop video mode as macOS and Windows do
(
&None,
&Some(Fullscreen::Exclusive(RootVideoMode {
video_mode: PlatformVideoMode::X(ref video_mode),
})),
)
| (
&Some(Fullscreen::Borderless(_)),
&Some(Fullscreen::Exclusive(RootVideoMode {
video_mode: PlatformVideoMode::X(ref video_mode),
})),
) => {
let monitor = video_mode.monitor.as_ref().unwrap();
shared_state_lock.desktop_video_mode =
Some((monitor.id, self.xconn.get_crtc_mode(monitor.id)));
}
// Restore desktop video mode upon exiting exclusive fullscreen
(&Some(Fullscreen::Exclusive(_)), &None)
| (&Some(Fullscreen::Exclusive(_)), &Some(Fullscreen::Borderless(_))) => {
let (monitor_id, mode_id) = shared_state_lock.desktop_video_mode.take().unwrap();
self.xconn
.set_crtc_config(monitor_id, mode_id)
.expect("failed to restore desktop video mode");
}
_ => (),
}
drop(shared_state_lock);
match fullscreen {
Move fullscreen modes to not touch physical resolutions (#270) * Fix X11 screen resolution change using XrandR The previous XF86 resolution switching was broken and everything seems to have moved on to xrandr. Use that instead while cleaning up the code a bit as well. * Use XRandR for actual multiscreen support in X11 * Use actual monitor names in X11 * Get rid of ptr::read usage in X11 * Use a bog standard Vec instead of VecDeque * Get rid of the XRandR mode switching stuff Wayland has made the decision that apps shouldn't change screen resolutions and just take the screens as they've been setup. In the modern world where GPU scaling is cheap and LCD panels are scaling anyway it makes no sense to make "physical" resolution changes when software should be taking care of it. This massively simplifies the code and makes it easier to extend to more niche setups like MST and videowalls. * Rename fullscreen options to match new semantics * Implement XRandR 1.5 support * Get rid of the FullScreen enum Moving to just having two states None and Some(MonitorId) and then being able to set full screen in the current monitor with something like: window.set_fullscreen(Some(window.current_monitor())); * Implement Window::get_current_monitor() Do it by iterating over the available monitors and finding which has the biggest overlap with the window. For this MonitorId needs a new get_position() that needs to be implemented for all platforms. * Add unimplemented get_position() to all MonitorId * Make get_current_monitor() platform specific * Add unimplemented get_current_monitor() to all * Implement proper primary monitor selection in X11 * Shut up some warnings * Remove libxxf86vm package from travis Since we're no longer using XF86 there's no need to keep the package around for CI. * Don't use new struct syntax * Fix indentation * Adjust Android/iOS fullscreen/maximized On Android and iOS we can assume single screen apps that are already fullscreen and maximized so there are a few methods that are implemented by just returning a fixed value or not doing anything. * Mark OSX/Win fullscreen/maximized unimplemented()! These would be safe as no-ops but we should make it explicit so there is more of an incentive to actually implement them.
2017-09-07 18:33:46 +10:00
None => {
2018-06-15 09:42:18 +10:00
let flusher = self.set_fullscreen_hint(false);
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
let mut shared_state_lock = self.shared_state.lock();
if let Some(position) = shared_state_lock.restore_position.take() {
2018-06-15 09:42:18 +10:00
self.set_position_inner(position.0, position.1).queue();
}
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
Some(flusher)
2019-06-25 02:14:55 +10:00
}
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
Some(fullscreen) => {
let (video_mode, monitor) = match fullscreen {
Fullscreen::Exclusive(RootVideoMode {
video_mode: PlatformVideoMode::X(ref video_mode),
}) => (Some(video_mode), video_mode.monitor.as_ref().unwrap()),
Fullscreen::Borderless(RootMonitorHandle {
inner: PlatformMonitorHandle::X(ref monitor),
}) => (None, monitor),
_ => unreachable!(),
};
// Don't set fullscreen on an invalid dummy monitor handle
if monitor.is_dummy() {
return None;
}
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
if let Some(video_mode) = video_mode {
// FIXME: this is actually not correct if we're setting the
// video mode to a resolution higher than the current
// desktop resolution, because XRandR does not automatically
// reposition the monitors to the right and below this
// monitor.
//
// What ends up happening is we will get the fullscreen
// window showing up on those monitors as well, because
// their virtual position now overlaps with the monitor that
// we just made larger..
//
// It'd be quite a bit of work to handle this correctly (and
// nobody else seems to bother doing this correctly either),
// so we're just leaving this broken. Fixing this would
// involve storing all CRTCs upon entering fullscreen,
// restoring them upon exit, and after entering fullscreen,
// repositioning displays to the right and below this
// display. I think there would still be edge cases that are
// difficult or impossible to handle correctly, e.g. what if
// a new monitor was plugged in while in fullscreen?
//
// I think we might just want to disallow setting the video
// mode higher than the current desktop video mode (I'm sure
// this will make someone unhappy, but it's very unusual for
// games to want to do this anyway).
self.xconn
.set_crtc_config(monitor.id, video_mode.native_mode)
.expect("failed to set video mode");
}
let window_position = self.outer_position_physical();
self.shared_state.lock().restore_position = Some(window_position);
let monitor_origin: (i32, i32) = monitor.position().into();
self.set_position_inner(monitor_origin.0, monitor_origin.1)
.queue();
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
Some(self.set_fullscreen_hint(true))
2019-06-25 02:14:55 +10:00
}
}
}
#[inline]
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
pub fn fullscreen(&self) -> Option<Fullscreen> {
let shared_state = self.shared_state.lock();
shared_state
.desired_fullscreen
.clone()
.unwrap_or_else(|| shared_state.fullscreen.clone())
}
2018-06-15 09:42:18 +10:00
#[inline]
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
pub fn set_fullscreen(&self, fullscreen: Option<Fullscreen>) {
if let Some(flusher) = self.set_fullscreen_inner(fullscreen) {
flusher
.sync()
Add exclusive fullscreen mode (#925) * Add exclusive fullscreen mode * Add `WindowExtMacOS::set_fullscreen_presentation_options` * Capture display for exclusive fullscreen on macOS * Fix applying video mode on macOS after a fullscreen cycle * Fix compilation on iOS * Set monitor appropriately for fullscreen on macOS * Fix exclusive to borderless fullscreen transitions on macOS * Fix borderless to exclusive fullscreen transition on macOS * Sort video modes on Windows * Fix fullscreen issues on Windows * Fix video mode changes during exclusive fullscreen on Windows * Add video mode sorting for macOS and iOS * Fix monitor `ns_screen` returning `None` after video mode change * Fix "multithreaded" example on macOS * Restore video mode upon closing an exclusive fullscreen window * Fix "multithreaded" example closing multiple windows at once * Fix compilation on Linux * Update FEATURES.md * Don't care about logical monitor groups on X11 * Add exclusive fullscreen for X11 * Update FEATURES.md * Fix transitions between exclusive and borderless fullscreen on X11 * Update CHANGELOG.md * Document that Wayland doesn't support exclusive fullscreen * Replace core-graphics display mode bindings on macOS * Use `panic!()` instead of `unreachable!()` in "fullscreen" example * Fix fullscreen "always on top" flag on Windows * Track current monitor for fullscreen in "multithreaded" example * Fix exclusive fullscreen sometimes not positioning window properly * Format * More formatting and fix CI issues * Fix formatting * Fix changelog formatting
2019-07-30 04:16:14 +10:00
.expect("Failed to change window fullscreen state");
self.invalidate_cached_frame_extents();
}
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
}
// Called by EventProcessor when a VisibilityNotify event is received
pub(crate) fn visibility_notify(&self) {
let mut shared_state = self.shared_state.lock();
match shared_state.visibility {
Visibility::No => unsafe {
(self.xconn.xlib.XUnmapWindow)(self.xconn.display, self.xwindow);
},
Visibility::Yes => (),
Visibility::YesWait => {
shared_state.visibility = Visibility::Yes;
if let Some(fullscreen) = shared_state.desired_fullscreen.take() {
drop(shared_state);
self.set_fullscreen(fullscreen);
}
}
}
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn current_monitor(&self) -> X11MonitorHandle {
self.shared_state.lock().last_monitor.clone()
Move fullscreen modes to not touch physical resolutions (#270) * Fix X11 screen resolution change using XrandR The previous XF86 resolution switching was broken and everything seems to have moved on to xrandr. Use that instead while cleaning up the code a bit as well. * Use XRandR for actual multiscreen support in X11 * Use actual monitor names in X11 * Get rid of ptr::read usage in X11 * Use a bog standard Vec instead of VecDeque * Get rid of the XRandR mode switching stuff Wayland has made the decision that apps shouldn't change screen resolutions and just take the screens as they've been setup. In the modern world where GPU scaling is cheap and LCD panels are scaling anyway it makes no sense to make "physical" resolution changes when software should be taking care of it. This massively simplifies the code and makes it easier to extend to more niche setups like MST and videowalls. * Rename fullscreen options to match new semantics * Implement XRandR 1.5 support * Get rid of the FullScreen enum Moving to just having two states None and Some(MonitorId) and then being able to set full screen in the current monitor with something like: window.set_fullscreen(Some(window.current_monitor())); * Implement Window::get_current_monitor() Do it by iterating over the available monitors and finding which has the biggest overlap with the window. For this MonitorId needs a new get_position() that needs to be implemented for all platforms. * Add unimplemented get_position() to all MonitorId * Make get_current_monitor() platform specific * Add unimplemented get_current_monitor() to all * Implement proper primary monitor selection in X11 * Shut up some warnings * Remove libxxf86vm package from travis Since we're no longer using XF86 there's no need to keep the package around for CI. * Don't use new struct syntax * Fix indentation * Adjust Android/iOS fullscreen/maximized On Android and iOS we can assume single screen apps that are already fullscreen and maximized so there are a few methods that are implemented by just returning a fixed value or not doing anything. * Mark OSX/Win fullscreen/maximized unimplemented()! These would be safe as no-ops but we should make it explicit so there is more of an incentive to actually implement them.
2017-09-07 18:33:46 +10:00
}
pub fn available_monitors(&self) -> Vec<X11MonitorHandle> {
self.xconn.available_monitors()
}
pub fn primary_monitor(&self) -> X11MonitorHandle {
self.xconn.primary_monitor()
}
fn set_minimized_inner(&self, minimized: bool) -> util::Flusher<'_> {
unsafe {
if minimized {
let screen = (self.xconn.xlib.XDefaultScreen)(self.xconn.display);
(self.xconn.xlib.XIconifyWindow)(self.xconn.display, self.xwindow, screen);
util::Flusher::new(&self.xconn)
} else {
let atom = self.xconn.get_atom_unchecked(b"_NET_ACTIVE_WINDOW\0");
self.xconn.send_client_msg(
self.xwindow,
self.root,
atom,
Some(ffi::SubstructureRedirectMask | ffi::SubstructureNotifyMask),
[1, ffi::CurrentTime as c_long, 0, 0, 0],
)
}
}
}
#[inline]
pub fn set_minimized(&self, minimized: bool) {
self.set_minimized_inner(minimized)
.flush()
.expect("Failed to change window minimization");
}
fn set_maximized_inner(&self, maximized: bool) -> util::Flusher<'_> {
let horz_atom = unsafe {
self.xconn
.get_atom_unchecked(b"_NET_WM_STATE_MAXIMIZED_HORZ\0")
};
let vert_atom = unsafe {
self.xconn
.get_atom_unchecked(b"_NET_WM_STATE_MAXIMIZED_VERT\0")
};
self.set_netwm(
maximized.into(),
(horz_atom as c_long, vert_atom as c_long, 0, 0),
)
}
2018-06-15 09:42:18 +10:00
#[inline]
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
pub fn set_maximized(&self, maximized: bool) {
self.set_maximized_inner(maximized)
.flush()
.expect("Failed to change window maximization");
self.invalidate_cached_frame_extents();
Move fullscreen modes to not touch physical resolutions (#270) * Fix X11 screen resolution change using XrandR The previous XF86 resolution switching was broken and everything seems to have moved on to xrandr. Use that instead while cleaning up the code a bit as well. * Use XRandR for actual multiscreen support in X11 * Use actual monitor names in X11 * Get rid of ptr::read usage in X11 * Use a bog standard Vec instead of VecDeque * Get rid of the XRandR mode switching stuff Wayland has made the decision that apps shouldn't change screen resolutions and just take the screens as they've been setup. In the modern world where GPU scaling is cheap and LCD panels are scaling anyway it makes no sense to make "physical" resolution changes when software should be taking care of it. This massively simplifies the code and makes it easier to extend to more niche setups like MST and videowalls. * Rename fullscreen options to match new semantics * Implement XRandR 1.5 support * Get rid of the FullScreen enum Moving to just having two states None and Some(MonitorId) and then being able to set full screen in the current monitor with something like: window.set_fullscreen(Some(window.current_monitor())); * Implement Window::get_current_monitor() Do it by iterating over the available monitors and finding which has the biggest overlap with the window. For this MonitorId needs a new get_position() that needs to be implemented for all platforms. * Add unimplemented get_position() to all MonitorId * Make get_current_monitor() platform specific * Add unimplemented get_current_monitor() to all * Implement proper primary monitor selection in X11 * Shut up some warnings * Remove libxxf86vm package from travis Since we're no longer using XF86 there's no need to keep the package around for CI. * Don't use new struct syntax * Fix indentation * Adjust Android/iOS fullscreen/maximized On Android and iOS we can assume single screen apps that are already fullscreen and maximized so there are a few methods that are implemented by just returning a fixed value or not doing anything. * Mark OSX/Win fullscreen/maximized unimplemented()! These would be safe as no-ops but we should make it explicit so there is more of an incentive to actually implement them.
2017-09-07 18:33:46 +10:00
}
fn set_title_inner(&self, title: &str) -> util::Flusher<'_> {
2018-05-29 21:48:47 +10:00
let wm_name_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_NAME\0") };
let utf8_atom = unsafe { self.xconn.get_atom_unchecked(b"UTF8_STRING\0") };
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
let title = CString::new(title).expect("Window title contained null byte");
unsafe {
2018-05-29 21:48:47 +10:00
(self.xconn.xlib.XStoreName)(
self.xconn.display,
self.xwindow,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
title.as_ptr() as *const c_char,
);
2018-05-29 21:48:47 +10:00
self.xconn.change_property(
self.xwindow,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
wm_name_atom,
utf8_atom,
util::PropMode::Replace,
title.as_bytes(),
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
)
}
}
2015-12-24 20:57:08 +11:00
2018-06-15 09:42:18 +10:00
#[inline]
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
pub fn set_title(&self, title: &str) {
self.set_title_inner(title)
.flush()
.expect("Failed to set window title");
}
fn set_decorations_inner(&self, decorations: bool) -> util::Flusher<'_> {
let mut hints = self.xconn.get_motif_hints(self.xwindow);
hints.set_decorations(decorations);
self.xconn.set_motif_hints(self.xwindow, &hints)
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
}
2018-06-15 09:42:18 +10:00
#[inline]
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
pub fn set_decorations(&self, decorations: bool) {
self.set_decorations_inner(decorations)
.flush()
.expect("Failed to set decoration state");
self.invalidate_cached_frame_extents();
}
fn set_maximizable_inner(&self, maximizable: bool) -> util::Flusher<'_> {
let mut hints = self.xconn.get_motif_hints(self.xwindow);
hints.set_maximizable(maximizable);
self.xconn.set_motif_hints(self.xwindow, &hints)
}
fn set_always_on_top_inner(&self, always_on_top: bool) -> util::Flusher<'_> {
2018-05-29 21:48:47 +10:00
let above_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_STATE_ABOVE\0") };
self.set_netwm(always_on_top.into(), (above_atom as c_long, 0, 0, 0))
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_always_on_top(&self, always_on_top: bool) {
self.set_always_on_top_inner(always_on_top)
.flush()
.expect("Failed to set always-on-top state");
}
fn set_icon_inner(&self, icon: Icon) -> util::Flusher<'_> {
2018-05-29 21:48:47 +10:00
let icon_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_ICON\0") };
2018-05-08 07:36:21 +10:00
let data = icon.to_cardinals();
2018-05-29 21:48:47 +10:00
self.xconn.change_property(
self.xwindow,
2018-05-27 22:49:35 +10:00
icon_atom,
ffi::XA_CARDINAL,
util::PropMode::Replace,
data.as_slice(),
)
2018-05-08 07:36:21 +10:00
}
fn unset_icon_inner(&self) -> util::Flusher<'_> {
2018-05-29 21:48:47 +10:00
let icon_atom = unsafe { self.xconn.get_atom_unchecked(b"_NET_WM_ICON\0") };
2018-05-08 07:36:21 +10:00
let empty_data: [util::Cardinal; 0] = [];
2018-05-29 21:48:47 +10:00
self.xconn.change_property(
self.xwindow,
2018-05-27 22:49:35 +10:00
icon_atom,
ffi::XA_CARDINAL,
util::PropMode::Replace,
&empty_data,
)
2018-05-08 07:36:21 +10:00
}
2018-06-15 09:42:18 +10:00
#[inline]
2018-05-08 07:36:21 +10:00
pub fn set_window_icon(&self, icon: Option<Icon>) {
match icon {
Some(icon) => self.set_icon_inner(icon),
None => self.unset_icon_inner(),
}
.flush()
.expect("Failed to set icons");
2018-05-08 07:36:21 +10:00
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_visible(&self, visible: bool) {
let mut shared_state = self.shared_state.lock();
match (visible, shared_state.visibility) {
(true, Visibility::Yes) | (true, Visibility::YesWait) | (false, Visibility::No) => {
return
}
_ => (),
}
if visible {
unsafe {
(self.xconn.xlib.XMapRaised)(self.xconn.display, self.xwindow);
}
self.xconn
.flush_requests()
.expect("Failed to call XMapRaised");
shared_state.visibility = Visibility::YesWait;
} else {
unsafe {
(self.xconn.xlib.XUnmapWindow)(self.xconn.display, self.xwindow);
}
self.xconn
.flush_requests()
.expect("Failed to call XUnmapWindow");
shared_state.visibility = Visibility::No;
}
}
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
fn update_cached_frame_extents(&self) {
let extents = self
.xconn
.get_frame_extents_heuristic(self.xwindow, self.root);
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
(*self.shared_state.lock()).frame_extents = Some(extents);
}
2018-06-15 09:42:18 +10:00
pub(crate) fn invalidate_cached_frame_extents(&self) {
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
(*self.shared_state.lock()).frame_extents.take();
}
pub(crate) fn outer_position_physical(&self) -> (i32, i32) {
2018-06-15 09:42:18 +10:00
let extents = (*self.shared_state.lock()).frame_extents.clone();
if let Some(extents) = extents {
let (x, y) = self.inner_position_physical();
extents.inner_pos_to_outer(x, y)
2018-06-15 09:42:18 +10:00
} else {
self.update_cached_frame_extents();
self.outer_position_physical()
2018-06-15 09:42:18 +10:00
}
}
2015-09-21 22:42:05 +10:00
#[inline]
pub fn outer_position(&self) -> Result<LogicalPosition, NotSupportedError> {
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
let extents = (*self.shared_state.lock()).frame_extents.clone();
if let Some(extents) = extents {
let logical = self.inner_position().unwrap();
Ok(extents.inner_pos_to_outer_logical(logical, self.hidpi_factor()))
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
} else {
self.update_cached_frame_extents();
self.outer_position()
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
}
}
pub(crate) fn inner_position_physical(&self) -> (i32, i32) {
// This should be okay to unwrap since the only error XTranslateCoordinates can return
// is BadWindow, and if the window handle is bad we have bigger problems.
self.xconn
.translate_coords(self.xwindow, self.root)
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
.map(|coords| (coords.x_rel_root, coords.y_rel_root))
.unwrap()
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn inner_position(&self) -> Result<LogicalPosition, NotSupportedError> {
Ok(self.logicalize_coords(self.inner_position_physical()))
2018-06-15 09:42:18 +10:00
}
pub(crate) fn set_position_inner(&self, mut x: i32, mut y: i32) -> util::Flusher<'_> {
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
// There are a few WMs that set client area position rather than window position, so
// we'll translate for consistency.
if util::wm_name_is_one_of(&["Enlightenment", "FVWM"]) {
let extents = (*self.shared_state.lock()).frame_extents.clone();
if let Some(extents) = extents {
x += extents.frame_extents.left as i32;
y += extents.frame_extents.top as i32;
} else {
self.update_cached_frame_extents();
2018-06-15 09:42:18 +10:00
return self.set_position_inner(x, y);
}
}
unsafe {
(self.xconn.xlib.XMoveWindow)(self.xconn.display, self.xwindow, x as c_int, y as c_int);
2018-06-15 09:42:18 +10:00
}
util::Flusher::new(&self.xconn)
}
pub(crate) fn set_position_physical(&self, x: i32, y: i32) {
self.set_position_inner(x, y)
.flush()
.expect("Failed to call `XMoveWindow`");
}
2015-09-21 22:42:05 +10:00
#[inline]
pub fn set_outer_position(&self, logical_position: LogicalPosition) {
let (x, y) = logical_position.to_physical(self.hidpi_factor()).into();
2018-06-15 09:42:18 +10:00
self.set_position_physical(x, y);
}
pub(crate) fn inner_size_physical(&self) -> (u32, u32) {
// This should be okay to unwrap since the only error XGetGeometry can return
// is BadWindow, and if the window handle is bad we have bigger problems.
self.xconn
.get_geometry(self.xwindow)
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
.map(|geo| (geo.width, geo.height))
.unwrap()
}
2015-09-21 22:42:05 +10:00
#[inline]
pub fn inner_size(&self) -> LogicalSize {
self.logicalize_size(self.inner_size_physical())
2018-06-15 09:42:18 +10:00
}
2015-09-21 22:42:05 +10:00
#[inline]
pub fn outer_size(&self) -> LogicalSize {
2018-06-15 09:42:18 +10:00
let extents = self.shared_state.lock().frame_extents.clone();
if let Some(extents) = extents {
let logical = self.inner_size();
extents.inner_size_to_outer_logical(logical, self.hidpi_factor())
2018-06-15 09:42:18 +10:00
} else {
self.update_cached_frame_extents();
self.outer_size()
2018-06-15 09:42:18 +10:00
}
}
pub(crate) fn set_inner_size_physical(&self, width: u32, height: u32) {
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
unsafe {
2018-05-29 21:48:47 +10:00
(self.xconn.xlib.XResizeWindow)(
self.xconn.display,
self.xwindow,
width as c_uint,
height as c_uint,
X11: General cleanup (#491) * X11: General cleanup This is almost entirely internal changes, and as usual, doesn't actually fix any problems people have complained about. - `XSetInputFocus` can't be called before the window is visible. This was previously handled by looping (with a sleep) and querying for the window's state until it was visible. Now we use `XIfEvent`, which blocks until we receive `VisibilityNotify`. Note that this can't be replaced with an `XSync` (I tried). - We now call `XSync` at the end of window creation and check for errors, assuring that broken windows are never returned. When creating invisible windows, this is the only time the output buffer is flushed during the entire window creation process (AFAIK). For visible windows, `XIfEvent` will generally flush, but window creation has overall been reduced to the minimum number of flushes. - `check_errors().expect()` has been a common pattern throughout the backend, but it seems that people (myself included) didn't make a distinction between using it after synchronous requests and asynchronous requests. Now we only use it after async requests if we flush first, though this still isn't correct (since the request likely hasn't been processed yet). The only real solution (besides forcing a sync *every time*) is to handle asynchronous errors *asynchronously*. For future work, I plan on adding logging, though I don't plan on actually *handling* those errors; that's more of something to hope for in the hypothetical async/await XCB paradise. - We now flush whenever it makes sense to. `util::Flusher` was added to force contributors to be aware of the output buffer. - `Window::get_position`, `Window::get_inner_position`, `Window::get_inner_size`, and `Window::get_outer_size` previously all required *several* round-trips. On my machine, it took an average of around 80µs. They've now been reduced to one round-trip each, which reduces my measurement to 16µs. This was accomplished simply by caching the frame extents, which are expensive to calculate (due to various queries and heuristics), but change infrequently and predictably. I still recommend that application developers use these methods sparingly and generally prefer storing the values from `Resized`/`Moved`, as that's zero overhead. - The above change enabled me to change the `Moved` event to supply window positions, rather than client area positions. Additionally, we no longer generate `Moved` for real (as in, not synthetic) `ConfigureNotify` events. Real `ConfigureNotify` events contain positions relative to the parent window, which are typically constant and useless. Since that position would be completely different from the root-relative positions supplied by synthetic `ConfigureNotify` events (which are the vast majority of them), that meant real `ConfigureNotify` events would *always* be detected as the position having changed, so the resultant `Moved` was multiple levels of misleading. In practice, this meant a garbage `Moved` would be sent every time the window was resized; now a resize has to actually change the window's position to be accompanied by `Moved`. - Every time we processed an `XI_Enter` event, we would leak 4 bytes via `util::query_pointer` (`XIQueryPointer`). `XIButtonState` contains a dynamically-allocated mask field which we weren't freeing. As this event occurs with fairly high frequency, long-running applications could easily accumulate substantial leaks. `util::PointerState::drop` now takes care of this. - The `util` module has been split up into several sub-modules, as it was getting rather lengthy. This accounts for a significant part of this diff, unfortunately. - Atoms are now cached. Xlib caches them too, so `XInternAtom` wouldn't typically be a round-trip anyway, but the added complexity is negligible. - Switched from `std::sync::Mutex` to `parking_lot::Mutex` (within this backend). There appears to be no downside to this, but if anyone finds one, this would be easy to revert. - The WM name and supported hints are now global to the application, and are updated upon `ReparentNotify`, which should detect when the WM was replaced (assuming a reparenting WM was involved, that is). Previously, these values were per-window and would never update, meaning replacing the WM could potentially lead to (admittedly very minor) problems. - The result of `Window2::create_empty_cursor` will now only be used if it actually succeeds. - `Window2::load_cursor` no longer re-allocates the cursor name. - `util::lookup_utf8` previously allocated a 16-byte buffer on the heap. Now it allocates a 1024-byte buffer on the stack, and falls back to dynamic allocation if the buffer is too small. This base buffer size is admittedly gratuitous, but less so if you're using IME. - `with_c_str` was finally removed. - Added `util::Format` enum to help prevent goofs when dealing with format arguments. - `util::get_property`, something I added way back in my first winit PR, only calculated offsets correctly for `util::Format::Char`. This was concealed by the accomodating buffer size, as it would be very rare for the offset to be needed; however, testing with a buffer size of 1, `util::Format::Long` would read from the same offset multiple times, and `util::Format::Short` would miss data. This function now works correctly for all formats, relying on the simple fact that the offset increases by the buffer size on each iteration. We also account for the extra byte that `XGetWindowProperty` allocates at the end of the buffer, and copy data from the buffer instead of moving it and taking ownership of the pointer. - Drag and drop now reliably works in release mode. This is presumably related to the `util::get_property` changes. - `util::change_property` now exists, which should make it easier to add features in the future. - The `EventsLoop` device map is no longer in a mutex. - `XConnection` now implements `Debug`. - Valgrind no longer complains about anything related to winit (with either the system allocator or jemalloc, though "not having valgrind complain about jemalloc" isn't something to strive for). * X11: Add better diagnostics when initialization fails * X11: Handle XIQueryDevice failure * X11: Use correct types in error handler
2018-05-03 23:15:49 +10:00
);
2018-05-29 21:48:47 +10:00
self.xconn.flush_requests()
}
.expect("Failed to call `XResizeWindow`");
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_inner_size(&self, logical_size: LogicalSize) {
let dpi_factor = self.hidpi_factor();
2018-06-15 09:42:18 +10:00
let (width, height) = logical_size.to_physical(dpi_factor).into();
self.set_inner_size_physical(width, height);
}
fn update_normal_hints<F>(&self, callback: F) -> Result<(), XError>
where
F: FnOnce(&mut util::NormalHints<'_>) -> (),
{
2018-06-15 09:42:18 +10:00
let mut normal_hints = self.xconn.get_normal_hints(self.xwindow)?;
callback(&mut normal_hints);
self.xconn
.set_normal_hints(self.xwindow, normal_hints)
.flush()
2018-06-15 09:42:18 +10:00
}
pub(crate) fn set_min_inner_size_physical(&self, dimensions: Option<(u32, u32)>) {
2018-06-15 09:42:18 +10:00
self.update_normal_hints(|normal_hints| normal_hints.set_min_size(dimensions))
.expect("Failed to call `XSetWMNormalHints`");
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_min_inner_size(&self, logical_dimensions: Option<LogicalSize>) {
self.shared_state.lock().min_inner_size = logical_dimensions;
let physical_dimensions = logical_dimensions
.map(|logical_dimensions| logical_dimensions.to_physical(self.hidpi_factor()).into());
self.set_min_inner_size_physical(physical_dimensions);
2018-06-15 09:42:18 +10:00
}
pub(crate) fn set_max_inner_size_physical(&self, dimensions: Option<(u32, u32)>) {
2018-06-15 09:42:18 +10:00
self.update_normal_hints(|normal_hints| normal_hints.set_max_size(dimensions))
.expect("Failed to call `XSetWMNormalHints`");
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_max_inner_size(&self, logical_dimensions: Option<LogicalSize>) {
self.shared_state.lock().max_inner_size = logical_dimensions;
let physical_dimensions = logical_dimensions
.map(|logical_dimensions| logical_dimensions.to_physical(self.hidpi_factor()).into());
self.set_max_inner_size_physical(physical_dimensions);
}
2018-06-15 09:42:18 +10:00
pub(crate) fn adjust_for_dpi(
&self,
old_dpi_factor: f64,
new_dpi_factor: f64,
width: f64,
height: f64,
) -> (f64, f64, util::Flusher<'_>) {
2018-06-15 09:42:18 +10:00
let scale_factor = new_dpi_factor / old_dpi_factor;
let new_width = width * scale_factor;
let new_height = height * scale_factor;
self.update_normal_hints(|normal_hints| {
let dpi_adjuster = |(width, height): (u32, u32)| -> (u32, u32) {
let new_width = width as f64 * scale_factor;
let new_height = height as f64 * scale_factor;
(new_width.round() as u32, new_height.round() as u32)
};
let max_size = normal_hints.get_max_size().map(&dpi_adjuster);
let min_size = normal_hints.get_min_size().map(&dpi_adjuster);
let resize_increments = normal_hints.get_resize_increments().map(&dpi_adjuster);
let base_size = normal_hints.get_base_size().map(&dpi_adjuster);
normal_hints.set_max_size(max_size);
normal_hints.set_min_size(min_size);
normal_hints.set_resize_increments(resize_increments);
normal_hints.set_base_size(base_size);
})
.expect("Failed to update normal hints");
unsafe {
2018-06-15 09:42:18 +10:00
(self.xconn.xlib.XResizeWindow)(
self.xconn.display,
self.xwindow,
new_width.round() as c_uint,
new_height.round() as c_uint,
);
}
(new_width, new_height, util::Flusher::new(&self.xconn))
}
pub fn set_resizable(&self, resizable: bool) {
if util::wm_name_is_one_of(&["Xfwm4"]) {
// Making the window unresizable on Xfwm prevents further changes to `WM_NORMAL_HINTS` from being detected.
// This makes it impossible for resizing to be re-enabled, and also breaks DPI scaling. As such, we choose
// the lesser of two evils and do nothing.
warn!("To avoid a WM bug, disabling resizing has no effect on Xfwm4");
return;
}
2018-06-15 09:42:18 +10:00
let (logical_min, logical_max) = if resizable {
let shared_state_lock = self.shared_state.lock();
(
shared_state_lock.min_inner_size,
shared_state_lock.max_inner_size,
)
} else {
let window_size = Some(self.inner_size());
2018-06-15 09:42:18 +10:00
(window_size.clone(), window_size)
};
self.set_maximizable_inner(resizable).queue();
let dpi_factor = self.hidpi_factor();
let min_inner_size = logical_min
2018-06-15 09:42:18 +10:00
.map(|logical_size| logical_size.to_physical(dpi_factor))
.map(Into::into);
let max_inner_size = logical_max
2018-06-15 09:42:18 +10:00
.map(|logical_size| logical_size.to_physical(dpi_factor))
.map(Into::into);
self.update_normal_hints(|normal_hints| {
normal_hints.set_min_size(min_inner_size);
normal_hints.set_max_size(max_inner_size);
})
.expect("Failed to call `XSetWMNormalHints`");
}
#[inline]
pub fn xlib_display(&self) -> *mut c_void {
2018-05-29 21:48:47 +10:00
self.xconn.display as _
}
2016-10-20 03:11:02 +11:00
#[inline]
pub fn xlib_screen_id(&self) -> c_int {
2018-05-29 21:48:47 +10:00
self.screen_id
2016-10-20 03:11:02 +11:00
}
#[inline]
pub fn xlib_xconnection(&self) -> Arc<XConnection> {
2018-05-29 21:48:47 +10:00
Arc::clone(&self.xconn)
2016-10-20 03:11:02 +11:00
}
#[inline]
pub fn xlib_window(&self) -> c_ulong {
2018-05-29 21:48:47 +10:00
self.xwindow
}
2015-09-21 22:42:05 +10:00
#[inline]
pub fn xcb_connection(&self) -> *mut c_void {
unsafe { (self.xconn.xlib_xcb.XGetXCBConnection)(self.xconn.display) as *mut _ }
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_cursor_icon(&self, cursor: CursorIcon) {
let old_cursor = replace(&mut *self.cursor.lock(), cursor);
if cursor != old_cursor && *self.cursor_visible.lock() {
self.xconn.set_cursor_icon(self.xwindow, Some(cursor));
}
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_cursor_grab(&self, grab: bool) -> Result<(), ExternalError> {
let mut grabbed_lock = self.cursor_grabbed.lock();
if grab == *grabbed_lock {
return Ok(());
}
unsafe {
// We ungrab before grabbing to prevent passive grabs from causing `AlreadyGrabbed`.
// Therefore, this is common to both codepaths.
(self.xconn.xlib.XUngrabPointer)(self.xconn.display, ffi::CurrentTime);
2016-03-22 05:42:54 +11:00
}
let result = if grab {
let result = unsafe {
(self.xconn.xlib.XGrabPointer)(
self.xconn.display,
self.xwindow,
ffi::True,
(ffi::ButtonPressMask
| ffi::ButtonReleaseMask
| ffi::EnterWindowMask
| ffi::LeaveWindowMask
| ffi::PointerMotionMask
| ffi::PointerMotionHintMask
| ffi::Button1MotionMask
| ffi::Button2MotionMask
| ffi::Button3MotionMask
| ffi::Button4MotionMask
| ffi::Button5MotionMask
| ffi::ButtonMotionMask
| ffi::KeymapStateMask) as c_uint,
ffi::GrabModeAsync,
ffi::GrabModeAsync,
self.xwindow,
0,
ffi::CurrentTime,
)
};
2016-03-22 05:42:54 +11:00
match result {
ffi::GrabSuccess => Ok(()),
ffi::AlreadyGrabbed => {
Err("Cursor could not be grabbed: already grabbed by another client")
2019-06-25 02:14:55 +10:00
}
ffi::GrabInvalidTime => Err("Cursor could not be grabbed: invalid time"),
ffi::GrabNotViewable => {
Err("Cursor could not be grabbed: grab location not viewable")
2019-06-25 02:14:55 +10:00
}
ffi::GrabFrozen => Err("Cursor could not be grabbed: frozen by another client"),
_ => unreachable!(),
}
.map_err(|err| ExternalError::Os(os_error!(OsError::XMisc(err))))
} else {
self.xconn
.flush_requests()
.map_err(|err| ExternalError::Os(os_error!(OsError::XError(err))))
};
if result.is_ok() {
*grabbed_lock = grab;
2016-03-22 05:42:54 +11:00
}
result
}
#[inline]
pub fn set_cursor_visible(&self, visible: bool) {
let mut visible_lock = self.cursor_visible.lock();
if visible == *visible_lock {
return;
}
let cursor = if visible {
Some(*self.cursor.lock())
} else {
None
};
*visible_lock = visible;
drop(visible_lock);
self.xconn.set_cursor_icon(self.xwindow, cursor);
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn hidpi_factor(&self) -> f64 {
self.current_monitor().hidpi_factor
}
pub fn set_cursor_position_physical(&self, x: i32, y: i32) -> Result<(), ExternalError> {
unsafe {
(self.xconn.xlib.XWarpPointer)(self.xconn.display, 0, self.xwindow, 0, 0, 0, 0, x, y);
self.xconn
.flush_requests()
.map_err(|e| ExternalError::Os(os_error!(OsError::XError(e))))
}
}
2018-06-15 09:42:18 +10:00
#[inline]
pub fn set_cursor_position(
&self,
logical_position: LogicalPosition,
) -> Result<(), ExternalError> {
let (x, y) = logical_position.to_physical(self.hidpi_factor()).into();
2018-06-15 09:42:18 +10:00
self.set_cursor_position_physical(x, y)
}
pub(crate) fn set_ime_position_physical(&self, x: i32, y: i32) {
let _ = self
.ime_sender
2018-06-15 09:42:18 +10:00
.lock()
.send((self.xwindow, x as i16, y as i16));
}
#[inline]
pub fn set_ime_position(&self, logical_spot: LogicalPosition) {
let (x, y) = logical_spot.to_physical(self.hidpi_factor()).into();
self.set_ime_position_physical(x, y);
2018-06-15 09:42:18 +10:00
}
#[inline]
pub fn id(&self) -> WindowId {
WindowId(self.xwindow)
}
#[inline]
pub fn request_redraw(&self) {
self.pending_redraws
.lock()
.unwrap()
.insert(WindowId(self.xwindow));
}
#[inline]
pub fn raw_window_handle(&self) -> XlibHandle {
XlibHandle {
window: self.xwindow,
display: self.xconn.display as _,
..XlibHandle::empty()
}
}
}