In some layouts this inconsistency results in crashes in
gdk_gl_texture_from_surface() since it uses gdk_gl_context_get_window() but
the returned window is not the same as the one that is being painted so
"window->current_paint.surface" is NULL. I saw this problem when packing a
GdkGLArea into a GtkPaned.
https://bugzilla.gnome.org/show_bug.cgi?id=743146
- Specifically request GL version when creating context. Just specifying core
profile bit results in the requested version defaulting to 1.0 which causes
the core profile bit to be ignored and an arbitrary compatability context to be
returned.
- Fix GL painting by removing GL calls that have been depricated by the 3.2 core
profile.
- Additionally remove glInvalidateFramebuffer() call, it is not supported by 3.2
core.
https://bugzilla.gnome.org/show_bug.cgi?id=742953
The ICCCM says:
If the specified property is None, the requestor is an obsolete client.
Owners are encouraged to support these clients by using the specified
target atom as the property name to be used for the reply.
Lets do that, instead of crashing.
https://bugzilla.gnome.org/show_bug.cgi?id=740613
The previous fix for this issue in 732af31424 was incomplete.
It seems that posix_fallocate gives an ENODEV error when
called on an fd opened with shm_open on freebsd. Fix up
the error check to only trigger if we get ENOSPC.
https://bugzilla.gnome.org/show_bug.cgi?id=742980
If we use GDK_GL_PROFILE_3_2_CORE we are asking for a core profile
according to the GLX_ARB_create_context_profile extension. For that,
we pass the GLX_CONTEXT_CORE_PROFILE_BIT_ARB value for the
GLX_CONTEXT_PROFILE_MASK_ARB attribute.
The specification for the extension says that:
If the requested OpenGL version is less than 3.2,
GLX_CONTEXT_PROFILE_MASK_ARB is ignored and the functionality
of the context is determined solely by the requested version.
Since we're asking for a core profile, we assume a GL version greater
than or equal to 3.2; thus, we don't need to specify the
GLX_CONTEXT_MAJOR_VERSION_ARB or the GLX_CONTEXT_MINOR_VERSION_ARB
attributes, and instead just rely on whatever version GLX gives us.
This seems to work around a strange issue in Mesa; if we ask for a core
profile and any version > 3.0, we get broken rendering on any shared
context we create.
Sending backingScaleFactor to a NULL NSWindow will silently give the
value 0 for the scale factor, causing insidious divide-by-zero bugs down
the line. This checks if the NSWindow is NULL first, as seems to happen
throughout the rest of the file.
Note that I don't have a hi-DPI OS X machine to test this on, though.
https://bugzilla.gnome.org/show_bug.cgi?id=738338
We've observed hangs of mutter when it initializes GTK+, which
are caused by initializing GL, which in turn makes xwayland
call back into mutter. With this change, mutter should just
disable GL support in GDK, and things will work.
This adds support for OpenGL to the GDK Windows backend using the WGL API
calls, which enables programs that uses the GTK+ GLArea widgets to work on
Windows as well.
This also adds a simple utility function to query for the version of OpenGL
that is supported by the Windows system, like the one provided by the X11
backend.
Many thanks to Alex (and Emmanuele, who started the OpenGL integration in
GTK+) who offered advice and help along the way, as well as the X11 and
Wayland backend for this work to refer to and to model upon.
https://bugzilla.gnome.org/show_bug.cgi?id=740795
As the alignments, strides and image formats may be different across
platforms, make the texture upload a vfunc to allow backends to override
the GL commands for uploading textures for the software implementation for
gdk_gl_texture_from_surface(), if necessary.
Suggested by Alex to avoid copying non-trivial portions of code which would
then add maintainenace burden.
https://bugzilla.gnome.org/show_bug.cgi?id=740795
We can't combine multiple draws into one for the software fallback,
because each quad has a different texture. And we generally don't
want to make a larger single texture because then we would have
to upload more data.
The ICCCM says:
If the specified property is None , the requestor is an obsolete client.
Owners are encouraged to support these clients by using the specified
target atom as the property name to be used for the reply.
Lets do that, instead of crashing.
https://bugzilla.gnome.org/show_bug.cgi?id=740613
NULL-plus-something could be seen by the compiler to attempt to do
arithmetic with void *, which is a GCCism. Instead, do the math normally
and cast the results as a void *.
https://bugzilla.gnome.org/show_bug.cgi?id=740605
Use xdg_surface_set_window_geometry() to tell the compositor about the
shadow widths, this makes some gnome-shell/mutter features (edge resistance,
frames around windows in the overview, side maximization, ...) work alright
with GTK+.
In order to add this, some other places in gdkwindow-wayland had to gain
some knowledge about margins:
- xdg_surface_configure() now syncs the shadow after applying the state,
and gdk_wayland_window_set_shadow_width() possibly reconfigures the
window in order to preserve window geometry. This is necessary to keep
shadows in sync with state/geometry changes, as this does not happen
all at once.
- xdg_popups relative to an xdg_surface are shown relative to buffer
coordinates, so the left/top margins must be added there.
https://bugzilla.gnome.org/show_bug.cgi?id=736742
This requires us to use GL_TRIANGLES and six verts per quad instead
of four, which makes me think it might not be worth it on
well-optimized GL drivers. However, from talking to some driver
developers about it, the GL_TRIANGLES should be faster, since this
means that there's one giant contiguous buffer instead of many small
buffers.
If we were really rendering a lot of quads, I'd use an element buffer
and GL_PRIMITIVE_RESTART, but we're really not ever rendering that
many quads, and the setup cost for that would just be too annoying.
It's unused. At the same time, rename "begin_paint_region" to
"begin_paint". This will help us clean up how GDK painting works
in the future to allow more creative use of double-buffering.
