| Age | Commit message (Collapse) | Author |
|---|
|
yasl/vulkan.h includes <vulkan/vulkan.h>. The C++ API test links the
shared library through shared_dep, which had only yasl's include path.
On macOS, Homebrew installs the Vulkan headers outside the default
search path, so the test did not compile.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
A drag that dropped on no YASL surface, such as one dropped on the
desktop or canceled, ended unheard: the public drag start set no end
listener. An application then waited for the drag's end forever.
The public layer now listens for the end. When no surface got the
drop, the origin gets DRAG with the new action YASL_DROP_CANCEL.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Weston 15 configures a minimized fullscreen window at 0 x 0 and then
rejects every frame with a protocol error. The test sent fullscreen,
maximize and minimize in one burst, so it failed whenever a frame landed
after that configure. It now waits for the window to enter and leave
fullscreen before it maximizes and minimizes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
A frame request now only records intent. Each loop step resolves it after
the callbacks that present, so a request made before the present waits for
that frame. An animation that asked first got two REFRESH events per frame
on X11 and with GL.
With no frame in flight, the REFRESH comes one refresh interval after the
last present or answer, so a request without a present cannot spin.
Queued events now go before an owed REFRESH. Input reaches the application
before it draws, and a queued REFRESH pays an owed one for the same surface.
The loop prepare only peeks at a lost event, so deliver can report it as a
final loss. A failed push from another thread wakes the loop.
Signals and posts run at the end of a step, after descriptors and timers.
A signal handler now sees the output that came with it.
A render-thread GL swap counts as in flight on Wayland.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
An application that has its own descriptors and timers needed a second
loop next to yasl's, and could not see yasl's key-repeat timer, clipboard
pipes or reconnect timer. Now there is one wait: the application owns a
yasl_loop, and yasl_init can run the display on it.
- Public yasl_loop_*: descriptors, timers and signals. Descriptor
callbacks return void. Timer IDs are 64 bits and never reused; a zero
delay runs in the next step. A nested step or run fails with EBUSY, a
step off the loop's thread with EPERM, and a quit made before a run ends
it at once. An interrupted wait keeps its deadline. Creating a loop
fails when it cannot make its wake pipe.
- The display drives the loop through prepare, wait and check hooks
instead of wrapping its step. Wayland finishes its read before any
callback runs, so loop callbacks may create surfaces and do roundtrips.
Readiness (queued events, an owed REFRESH) is derived in the step, so
the wait never blocks over pending work. A draw that cannot proceed for
want of a buffer is no longer counted as work.
- yasl_init takes struct yasl_options. With loop and on_event, each step
delivers events to on_event and yasl_wait_event fails with EBUSY.
EALREADY only for the same options, EBUSY for others.
- yasl_surface_request_frame: a REFRESH once the window can show a new
frame, for pacing animation.
- DISPLAY_LOST reports whether the display reconnects. Pull waits still
fail with ENOTCONN after a final loss.
- Events pushed from another thread wake the loop.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Add yasl_cpu_present_damage to submit partial damage rectangles.
Skip commit timers for CPU surfaces to prevent display delay.
Remove frame callback gating in surface_draw_frame when free buffers exist.
Flush the display immediately on CPU present.
Forward tearing option to window specifications.
|
|
This reverts commit b2f963670845e7e1bcb9f3b141e12ea7ba7560a4.
|
|
The caret's logical rectangle depends on the surface scale. A scale
change left the compositor with the old rectangle until the app sent a
new caret. Send the caret again on every rescale while text input is
active.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
The Wayland backend passed yasl's purpose values straight to
set_content_type. The values differ: yasl TERMINAL (3) reached the
compositor as the protocol's NUMBER. Map each yasl purpose to its
protocol constant. Unknown values fall back to normal text.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
On a real compositor, input from the user and focus-stealing
prevention reach the test window. Drain real events before checking
the order of pushed ones, check only that gaining focus does not reset
input, and accept ESTALE for a primary selection offered while the new
window lacks keyboard focus.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Integration tests now expect ESTALE from move, resize, window menu,
drag and popups when nothing presses a button, and the X11 input test
expects the same from the window menu and move.
Resolving the previous commit's merge also made the direct SHM path
check the window's own visual (possibly 32-bit ARGB) and dropped the
held-button check from place_under_pointer, which yasl.h never asked
for.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
- Losing the X server queues DISPLAY_LOST instead of exiting. A process-wide
IO error handler returns for YASL displays, so Xlib calls the per-display
exit handler and returns; later waits fail with ENOTCONN, and dispatch and
flush make no Xlib calls on a lost display.
- Errors on other displays chain to the previous handler, restored when the
last YASL display closes.
- MIT-SHM attach is synced; a refused attach detaches and falls back to
XPutImage. Shared images are used only for 32-bit TrueColor 0xAARRGGBB.
- Undecorated windows set MWM_HINTS_DECORATIONS and leave functions alone.
- Fixed-size windows keep their min/max hints in step with set_size.
- Swap interval is stored per surface and applied when it becomes current;
gl_make_current(NULL) without EGL succeeds as a no-op.
- XInitThreads runs before the first connection.
- Move, resize, window menu, drag start and place-under-pointer need a held
pointer button (ESTALE). Move and resize also need _NET_WM_MOVERESIZE on
the root (ENOTSUP) and send the held button.
- _GTK_SHOW_WINDOW_MENU carries the XI2 client pointer and root coordinates.
- Popups always present through CPU and grab the pointer (XI2 device grab
when available); a press outside every YASL window closes the topmost popup.
- Drag start while active returns EBUSY; drag carry without a drag EINVAL.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
- Wayland: a touch contact is a press for move, resize, menu and drag
again, and a compositor without xdg-system-bell reports ENOTSUP.
- yasl_surface_set_cursor no longer reports a stale errno.
- The X11 input test expects the TIMESTAMP and MULTIPLE targets and
drives an INCR transfer instead of expecting INCR to be rejected.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
- GL swaps on a rendering thread no longer touch frame callbacks or
presentation feedback. They set a per-surface flag and wake the loop; the
owner thread arms both at its next flush and commits, so the frame callback
fires without waiting for the application's next swap.
- Clear the offline flag only after the reconnect rebuild succeeds, and not
when the connection was lost again during it.
- make_current fails with ENOTCONN for a window without a context. Detaching
is a no-op when nothing is current or the EGL display is gone.
- Window move, resize, window menu and drag start need a pointer button or
tablet tip that is still held; otherwise they fail with ESTALE and use that
press's serial. Popups fail with ESTALE without an input serial and with
ENOTCONN while the display is lost.
- An oversized device size shrinks to the buffer limits, keeping the aspect
ratio, instead of closing the window.
- FIXED_SIZE windows pin xdg min and max on every set_size. The 64px minimum
applies only to compositor-suggested sizes, not to application sizes.
- The keyboard-enter serial becomes the input serial. Clipboard copy with no
keyboard focus fails with ESTALE.
- Relative-pointer windows relock after a reconnect rebuild.
- errno: EBUSY for an active drag, ENOTSUP for place_under_pointer and for GL
without EGL, EINVAL for drag carry without a drag.
- Default window title and app ID are "YASL".
- Pointer motion and button events carry unrounded device-pixel positions in
precise_x/precise_y with precise_position set.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
Clipboard and primary selection:
- Take ownership at the time of the latest key or button event instead of
CurrentTime; TIMESTAMP answers with the time each selection was taken.
- Forget the stored text on SelectionClear.
- Answer MULTIPLE (failed pairs get None) and refuse selections other than
CLIPBOARD and PRIMARY.
- Send payloads above the request limit with INCR instead of refusing them,
with a per-chunk timeout and cleanup on display close.
- Receive INCR transfers: delete the marker to start, append chunks on
PropertyNewValue, and finish on a zero-length chunk.
Translucency:
- CPU toplevels and popups use a 32-bit TrueColor visual with their own
colormap when the server has one. Shared-memory and plain images use the
window's visual and depth.
HiDPI:
- Read Xft.dpi at display open and derive the scale, clamped to 1x-4x.
Logical creation sizes and set_size requests become device pixels, and
reported geometry carries the scale.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
Add tests/common/test_public_api.c: before yasl_init, argument errors are reported first, owner-thread calls fail with ENODEV, queries return their documented fallbacks, yasl_drop_next_path parses a crafted text/uri-list, yasl_get_system_string follows its length and NUL rules, and yasl_wake is a no-op with no display.
Build tests/test_header_c11.c as strict C11 and C17 with -pedantic-errors.
Extend the surface integration test, run on weston and Xwayland, with title, window requests, bell, desktop queries, drag cancel, input-less move/menu/drag calls, a popup (its parent's destroy must also remove it, checked through the display's surface list), and a clipboard round trip.
Run tests/x11/test_input.c in the X11 phase of run_surfaces.py through an optional trailing argument.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
- Undecorated windows can become key and main; popups never take focus.
- Undecorated windows stay miniaturizable and resizable unless FIXED_SIZE.
- The close button sends a CLOSE request (windowShouldClose:) and does not
close; the app destroys the surface to comply.
- Refresh runs during live resize: the modal flag is set between
windowWillStartLiveResize: and windowDidEndLiveResize:, and CPU surfaces
refresh before the frame is rendered. A destroyed window clears the flag.
- presentsWithTransaction is only changed for CPU presenters.
- Display open and surface creation return EPERM off the main thread.
- Popups are always CPU surfaces with a software queue, also under a Metal
parent.
- Interactive move and drag start use the recorded mouse-down; without a
press they set ESTALE. Interactive resize and window menu return ENOTSUP.
- Notifications: ENOTSUP outside an app bundle; one authorization request,
notifications queued until it is granted, EACCES after a denial.
- Titles use the full string instead of a 159-byte copy that could split a
UTF-8 sequence.
- finishLaunching runs once per process, not once per display open.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
|
|
yasl.h now holds everything except Vulkan: 45 functions instead of 62,
and three installed headers instead of ten.
- One constructor: yasl_surface_create takes CPU, GL or NATIVE, with
zero-initialized GL options meaning 3.2 core, RGBA8, D24S8.
- Events carry action, mods, x, y and text at the top level; pen
positions are now logical like pointer positions.
- yasl_cpu_pixels/yasl_cpu_present draw into one persistent canvas,
cutting three full-frame copies per frame.
- yasl_surface_get_info, yasl_surface_request, YASL_CURSOR_LOCKED,
yasl_surface_set_text_input, yasl_surface_begin_move_resize and
yasl_surface_get_native replace groups of narrower calls.
- Interactive calls drop the decorative trigger event argument.
- yasl_wait_event returns 0 only on timeout or yasl_wake, the refresh
callback runs only in native modal loops, focus loss sends
INPUT_RESET, and a lost event no longer fails every later wait.
- Unsupported operations and bad arguments report ENOTSUP and EINVAL
instead of EIO; the keycodes header no longer needs C23.
- The shared library gets a versioned SONAME and version macros.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
|
|
|
|
|
|
|
|
Windows now outlive the compositor. On connection loss YASL sends
DISPLAY_LOST instead of closing every window, retries the connection on
a loop timer, rebuilds each toplevel with its title, size, state and GL
context, and sends DISPLAY_RESTORED. CPU windows keep their pixels;
GPU renderers drop Vulkan surfaces and GL objects on loss and rebuild
them on restore.
- Free the dead connection's objects on the next wait, after the
application has released GPU objects built on them.
- Report a lost connection as an event rather than a failed wait.
- Refuse window and Vulkan surface creation while offline with the new
YASL_ERROR_DISCONNECTED.
- Close popups, cancel drags and fail pending clipboard requests on loss.
- Print only protocol errors, not ordinary disconnects.
- Examples survive a restart; game.c rebinds its GL context.
- Add an end-to-end test that kills Weston under CPU, GL and Vulkan
windows and requires them to recover.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Applications now need far less ceremony: hello.c is a 20-line loop that
handles REFRESH, CLOSE and QUIT and draws into the surface's own pixels.
- Fold GEOMETRY, STATE and FOCUS into one WINDOW event carrying size,
scale and state; unchanged values send nothing.
- Guarantee REFRESH once a window has a size and after every resize,
never at size zero.
- Merge DROP_HOVER into DROP with over/leave/data actions and give drag
and drop actions enums.
- Keep event strings valid until the next wait and drop
yasl_event_release; empty or failed waits report NONE.
- Replace copy-in presentation with a lent pixel buffer:
yasl_cpu_pixels and yasl_cpu_present.
- Remove yasl_poll_event, yasl_cpu_read_pixels, unused event fields and
comments that restated names.
- Wayland: resize floating windows in request_size instead of reporting
it unsupported, and stop marking new GPU surfaces as Vulkan so
yasl_vulkan_create_surface no longer fails with NATIVE_WINDOW_IN_USE.
Also includes the in-progress public API refactor that was uncommitted
in the tree (removal of the app/loop layer, new surface, GL and Vulkan
entry points).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
|
|
Each of the six audit requests now lists its commit and the contract
that wsap_term asked to have defined. Cocoa is still listed as not
built or run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
With LTO, GCC inlines the IME handlers into the test that passes a
32-byte buffer and warns that strndup's 256 KiB bound exceeds it
(-Wstringop-overread). The read never passed the NUL, since libwayland
always terminates protocol strings, but the warning reached consumers
such as wsap_term.
The text is now measured with strlen, capped at IME_TEXT_MAX, and copied,
which keeps the same truncation. An LTO release build is warning-free.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Drops arrive as a CRLF-separated text/uri-list of file URLs, so every
consumer percent-decodes them; wsap_term does it in its own drop code.
yasl_drop_next_path() walks the list and returns one decoded local path
at a time. It accepts file:///, file:/, and file://localhost or this
machine's name, ends the path at '?' or '#', and skips comments, other
schemes, remote hosts, malformed escapes, and escaped NUL bytes. A buffer
that is too small leaves the cursor in place, like the other functions
that fill a buffer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
wsap_term writes PNG screenshots for its debug server and keeps its own
copy of the frame because YASL had no readback.
yasl_surface_read_pixels() copies the frame from the last successful
present, or reports its extent when the buffer is too small. X11 and
Cocoa keep it in their single CPU target until the next draw starts;
X11 now also forgets it when a resize replaces the target, and Cocoa
records the presented extent because a resize can change the size before
the next draw. Wayland remembers which buffer it committed and forgets it
when a draw reuses that buffer.
The backend test checks the copy on Wayland and X11, and that draws that
never present do not show through. The Cocoa path is not built or run
here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
wsap_term hardcodes 400 ms in three places to count clicks.
yasl_display_double_click_ms() reads the interval from the system. X11
parses Net/DoubleClickTime from the XSETTINGS owner's property. Wayland
has no protocol for it, so it asks the Settings portal for KDE's
DoubleClickInterval, which KDE sends as a string and which exists only
when the user set it, then GNOME's double-click. X11 without XSETTINGS
asks the portal too. Cocoa reads NSEvent.doubleClickInterval. Missing or
implausible values give YASL_DOUBLE_CLICK_MS_DEFAULT, 400 ms.
The X11 state test covers the XSETTINGS parser in both byte orders and
with truncation. The Xwayland test owns the settings selection. The
desktop test serves both portal keys. The Cocoa path is not built or run
here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
wsap_term keeps per-platform code for opening links, notifications, the
bell, high-contrast detection, the executable and bundle paths, and the
default locale. These now live behind <yasl/desktop.h>.
On Linux, src/platform/freedesktop serves both backends. Opening a URI
runs xdg-open from PATH through a double fork, so the opener gets default
signal dispositions, an empty mask, a new session, and none of the
application's descriptors. Notifications and the Settings portal use the
session bus through libdbus-1, an optional dependency behind -Ddbus.
Appearance falls back to GTK_THEME without a portal. The bell uses
xdg-system-bell on Wayland and the XKB bell on X11.
Cocoa implements the same calls with NSWorkspace, UserNotifications, the
effective appearance, NSBeep, the bundle, and CFLocale. That code is not
built or run here; it needs a macOS check.
The desktop test runs under dbus-run-session against a fake notification
server and portal, and uses a fake xdg-open to check argument passing,
signal state, and descriptor inheritance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Generated with wayland-scanner 1.26.0 from wayland-protocols' staging
xdg-system-bell-v1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Pointer events set button only on press and release, so wsap_term tracked
held buttons itself to tell a drag from a hover.
yasl_pointer_event.buttons now holds a YASL_BUTTON_BIT() mask of the buttons
held after each event. The common input queue tracks it from the DOWN and
UP events it queues, and clears it on INPUT_RESET and on allocation
failure. Wayland also clears it on leave, since the compositor sends leave
during a press only when it takes the pointer and no release follows. X11
and Cocoa keep it, because they deliver motion and the release to the
pressed window after leave.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
wsap_term runs its own eventfds to wake the loop from its pty worker and
accessibility threads, and its own signalfd or self-pipe per platform.
yasl_loop_kick() is now documented as safe from any thread and from a
signal handler. yasl_loop_post() queues a function to run on the owning
thread during a later step. yasl_loop_watch_signals() installs a
process-wide SA_RESTART handler per signal that records the signal and
kicks the loop, so callbacks run on the loop thread on every platform
without touching thread masks.
Cocoa dispatch waited only for AppKit events, so kicks, posts, and fd
watches could not end an indefinite wait, and yasl_post_event() did not
wake yasl_run(). While waiting, Cocoa now mirrors the loop's descriptors
in a kqueue whose run loop source posts a wake event. If kqueue rejects a
descriptor, the wait is bounded to 10 ms.
The Cocoa change is not built or run here; it needs a macOS check.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Drag targets, drag images, and X11 file drops are now implemented, so items
1-3 are resolved. Item 9 moves to the paths the audit reports. The new
requests from the wsap_term audit are listed as open.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
The README links to examples/hello.c, so a copy of its draw callback
only goes stale.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Red grows to the right, green downward, and blue fades as red grows.
Rows now step by the frame's stride_bytes instead of assuming packed
pixels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Without yasl_app_desc.event, a close request had nowhere to go and the
window stayed open. The runner now stops instead, so hello drops its
event callback.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
The game assumed yasl_surface_set_pointer_mode() always succeeded, so on
backends without relative mode it believed the pointer was captured.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
yasl_surface_set_pointer_mode(YASL_POINTER_RELATIVE) now takes a
persistent zwp_locked_pointer_v1 on the surface and hides its cursor.
While the compositor keeps the lock engaged, zwp_relative_pointer_v1
motion is pushed as POINTER_MOTION with pos frozen and delta in device
pixels. When the lock disengages, as on focus loss, motion is absolute
again with a delta derived by the input queue.
The lock follows the pointer's seat: it moves to another seat when its
seat loses the pointer, and is released on NORMAL and on surface
destruction. Compositors without both protocols return false.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Generated with wayland-scanner 1.26.0 from wayland-protocols' unstable
relative-pointer-unstable-v1 and pointer-constraints-unstable-v1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
hello no longer needs an init callback: main passes its window spec
through yasl_app_desc.window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
A new optional yasl_app_desc.window spec makes yasl_run() create that
toplevel and request its first frame before init(), which is now
optional too. If creation fails, yasl_run() closes the display and
returns an error without calling init().
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
hello only tracked a failure flag, so keep it as a file-scope static like
simple_window and leave yasl_app_desc.userdata unset.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
Port backend, input, tablet and app tests from yasl_input_event and
YASL_INPUT_* to struct yasl_event and YASL_EVENT_*. The input allocation
test links input.c on its own, so it stubs yasl_loop_now_ns() for the new
event timestamps.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
The core input queue and the yasl_run() runner each had their own event
type, and the runner translated one into the other. Both now share
struct yasl_event in <yasl/yasl.h>: yasl_surface_poll_event() returns the
same events the app event callback receives, so the runner forwards them
unchanged.
Pointer input becomes POINTER_DOWN/UP/MOTION/SCROLL/LEAVE with a named
yasl_pointer_button and a motion delta computed in the input queue. TEXT,
PASTE and DROP share one text payload, scale changes fold into
WINDOW_RESIZE, and every event carries a monotonic time_ns.
yasl_surface_set_mouse_mode() is replaced by
yasl_surface_set_pointer_mode(); no backend implements relative mode yet.
The export check now only counts YASL_API declarations, so function
pointer fields returning enum yasl_app_result are not mistaken for
exports.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
|
|
|
|
|
|
|
