Much of libdecor is initialized only after certain events have been
received from the compositor and some parts of libdecor 0.1 are unsafe
to use until this delayed initialization has completed.
Since libdecor does not provide an API to query if or be notified when
this has happened, GLFW processed events until its newly created
libdecor frame had created its XDG shell objects.
This commit switches to using a generic Wayland sync point created just
after libdecor (and presumably its plugin) has set up its delayed
initialization, instead of relying on the more specific implementation
detail mentioned above.
It also makes this wait mandatory before the first libdecor frame is
created instead of a pre-condition for certain libdecor frame calls,
hopefully removing even more dependence on implementation details.
The bitmask passed to MsgWaitForMultipleObjects was missing
QS_SENDMESSAGE, causing glfwWaitEventsTimeout not to return when the
thread received messages sent from other threads.
Fixes#2408
A GLFW_CURSOR_UNAVAILABLE error would be emitted each time the cursor
moved over the fallback decorations if the standard cursor shape
appropriate for that part was missing on the system.
These errors served no useful purpose and have been removed.
This replaces the previous scheme where scancodes were equal to keycodes
(an implausible and potentially misleading situation) with a set of
scancodes invented purely for the null platform.
This is partly based on the implementation of libdecor support in
PR #1693 by @ christianrauch.
Where available, the libdecor library is loaded at init and becomes the
preferred method for window decorations. On compositors that support
XDG decorations, libdecor in turn uses those. If not, libdecor has
a plug-in archtecture and may load additional libraries to either use
compositor-specific decorations or draw its own.
If necessary, support for libdecor can be disabled with the
GLFW_WAYLAND_LIBDECOR init hint. This is mostly in case some part of
the dynamic loading or duplication of header material added here turns
out to cause problems with future versions of libdecor-0.so.0.
Fixes#1639Closes#1693
Related to #1725
This commit simplifies the detection of which element of a fallback
decorated window the pointer has entered. Instead of looping through
the list of windows, the user pointer of fallback decoration surfaces is
set to the GLFW window object.
This replaces (one case of) manual management of weak links between
windows and monitors, both objects with complex life times, with
wl_object pointers used as opaque key values.
This is in preparation for adding support for libdecor, which creates
its own proxies on our display. It will likely also be helpful to some
people using native access on Wayland.
This is partly based on the implementation of libdecor support in
PR #1693 by @ christianrauch.
If a window was initially fullscreen then it would not get an XDG
decoration object. If the window was later switched to windowed mode it
would then get fallback decorations instead of XDG ones.
Because EGL ties client API support to configs, attempts to create
a context with an unavailable client API will fail with the unhelpful
"failed to find suitable config" error description.
This attempts to detect cases where there are usable configs for the
other client API and emit a hopefully more helpful error.
Related to #2173
We only care about displayable pixel formats (as defined in
WGL_ARB_pixel_format) for window context creation.
This changes pixel format enumeration to ignore non-displayable formats,
which are specified to be listed after displayable ones, by always using
the return value of DescribePixelFormat as the pixel format count.
On systems lacking the EGL_EXT_present_opaque extension, some
compositors treat any buffer with an alpha channel as per-pixel
transparent.
This commit ignores any EGLConfig with an alpha channel if the extension
is missing and the window is created with GLFW_TRANSPARENT_FRAMEBUFFER
set to false.
This is technically not a breaking change since GLFW_ALPHA_BITS is not
a hard constraint, but it is still going to inconvenience anyone using
the framebuffer alpa channel to store other kinds of data.
Related to #1895
This adds window hints for the initial position, in screen coordinates,
of a window. The special value GLFW_ANY_POSITION means the window
manager will be allowed to position the window.
It is not possible to set window positions on Wayland and GLFW will
always behave as if these hints are set to GLFW_ANY_POSITION.
Fixes#1603Fixes#1747
This adds a window hint string for the xdg_toplevel::app_id, which is
used by desktop environments to connect windows with application icons
and other information. This is similar to the WM_CLASS property on X11.
A few very minor fixes were done by @elmindreda during merge.
Fixes#2121Closes#2122
The cursor theme was only loaded if the chosen seat had a mouse
(wl_pointer) during initialization. If a mouse was connected only after
glfwInit, there would be no cursor theme but the rest of the cursor
related code assumed one had already been loaded.
This also moves the details of cursor theme loading out into a separate
function to declutter platform init.
Because the original cursor theme loading code checked whether we got
a wl_shm, and because the rest of the code just assumes we have
a wl_shm, initialization will now fail if there isn't one.
Fixes#1450
If the key or character callback performs actions that indirectly
updates the key repeat timer, those changes would be undone once the key
callback returned.
This fixes the order of operations so that key repeat is fully set up
before the key related events are emitted.
This fixes a regression introduced by
2c204ab52e. This broke the check for
whether there is already a GLFW joystick object for a given input
device, making it always fail.
The NSWindowCollectionBehaviorFullScreenNone enum value is missing from
system headers on 10.11 despite the documentation claiming it was added
in 10.7.
Unfortunately Apple has taken down all API release notes for versions
prior to 10.14.
This fixes a build failure introduced with
98d6e8485b.
According to both Apple and LunarG, a private copy of the macOS Vulkan
loader libvulkan.1.dylib should be placed in the Frameworks directory
of the bundle and not its main executable directory.
This commit updates the dynamic loading path accordingly.
Fixes#2113Closes#2120
Whenever GLFW changed the window style mask, a new mask was created
from scratch based on the attributes set on the GLFW window object.
This caused us to potentially clear unrelated window style bits.
This was always wrong but became a critical issue when Cocoa began
throwing an exception if an application cleared the
NSWindowStyleMaskFullScreen while the window is in macOS fullscreen.
This commit reworks all style mask editing so it only changes the
relevant bits, preserving all others.
This is only a narrow bug fix to prevent crashes, intended for the
stable branch. Our interaction with macOS fullscreen is still very
poor. The next step after this is a set of patches that improve the
interaction between the current API and macOS fullscreen.
Fixes#1886Fixes#2110
The reasoning here is that glfwRestoreWindow will change nothing for
a windowed non-resizable window on Cocoa, and silently refusing to
maximize seems slightly more like something other platforms would do.
This is possibly either the right thing to do or the wrong one.
If a fullscreen window with GLFW_DECORATED set had its XDG decorations
changed to client mode by the compositor, it would seemingly receive
GLFW fallback decorations as if it was windowed mode.
This is possibly related to #2001.
Note that the handling of configure events, acks and commits is still
not ideal. This is just a small step in, hopefully, a good direction.
Fullscreen toggling via glfwSetWindowMonitor now works on Weston, but
mostly incidentally.
If the xdg_toplevel has a decoration, we need to wait for its first
configure event as well before we are allowed to attach the first
buffer.
It seems racy to assume that this will always happen inside the first
surface configure sequence, so this commit makes that condition
explicit. This may turn out to have been overly defensive.
Refer to the XDG decoration mode (or the lack of one) directly instead
of setting a boolean in a struct meant for the fallback decorations.
This makes things a bit more verbose but is in preparation for
a refactoring of all decoration paths.
When showing a window that had already been shown once (and so already
had its shell objects), GLFW would attach a new buffer and commit it
before waiting for the next configure event. This was a violation of
the XDG shell protocol.
This was allowed to work as intended on GNOME and KDE without error.
However wlroots based compositors would (correctly) emit an error.
Unfortunately, I haven't been able to find a way to get both KDE, GNOME
and Sway to send the configure event we need in order to map the
wl_surface again while keeping our existing shell objects, so with this
commit we now create them for each call to glfwShowWindow and destroy
them for each call to glfwHideWindow.
Fixes#1268
If a window was created as maximized, or created as hidden and then
iconified or maximized before first being shown, that state was lost and
the window was shown as restored.
Window iconfication and maximization events were being emitted before
xdg_surface::configure, making it possible for user code to indirectly
commit surface changes from those event callbacks before
xdg_surface::ack_configure.
This postpones those events until after the ack has been sent.
Content scale events would be emitted when a window surface entered or
left an output, but not when one of a window's current outputs had its
scale changed.
GLFW would report a monitor as connected each time its wl_output
received an update, for example if its scale changed.
This would also cause the monitor to be added to the monitor array
again, causing glfwTerminate to segfault when it attempted to destroy
its already destroyed wl_output.
This is a temporary local fix to have updates to GLFW_DECORATED mostly
work as intended. The whole decoration state machine needs to be
restructured, but not by this commit.
