It is also possible to enable F16C instructions on 32-bit Visual Studio builds,
so also check for the compiler's ability to build F16C code on Visual Studio
32-bit instead of just bailing out.
We need to use __cpuid() to check for the presence of F16C instructions on
Visual Studio builds, and call the half_to_float4() or float_to_half4()
implementation accordingly, as the __builtin_cpu...() functions are strictly
for GCC or CLang only.
Also, since __m128i_u is not a standard intrisics type across the board, just
use __m128i on Visual Studio as it is safe to do so there for use for
_mm_loadl_epi64().
Like running on Darwin, we cannot use the alias __attribute__ as __attribute__
is also for GCC and CLang only.
The __builtin_cpu...() intrisics are strictly for GCC/CLang-based compilers, so
don't use them in the checking code on Visual Studio. The test code will still
compile without this change, but will certainly fail if we want it to link.
Propagate the focus-on-click setting to the button
inside, so that setting menubuttons as !focus-on-click
works as expected. This helps for menubuttons in
header bars, where dragging on the button will otherwise
steal focus from the content.
When 9-slicing shadows, omit the center tile when it is
entirely contained in the outline (that is not always
the case, depending on corners and offsets).
gsk_rounded_rect_contains_rect was calling
gsk_rounded_rect_contains_point, which potentially
checks all four corners, for a total of up to 16
corner/point checks. But there is no need to do
more than 4 such checks to answer the question.
Commit 3dbf5038fa tried to defer focus changes
until after rendering is done. But it failed to do so, since
the toplevel ::render handler is still before rendering of
popups that are attached to that toplevel. To do this
properly, we need to do it in the AFTER_PAINT frame clock
phase.
Fixes: #3725
We used to override cursor to use all-scroll while the
content is being scrolled. Unfortunately, there is several
problems with this:
- It is really only expected certain devices, and we don't
have the device information on Wayland
- With the way cursor setting works in GTK4, non-NULL cursors
of the content (eg the text views ibeam) win, making the
scroll cursor not show up
- Under X11, we seem to miss scroll end events and then
the scroll cursor gets stuck
Therefore, just remove this feature.
Opportunistically use the coloring program for
drawing underlines instead of the color program.
This avoids program changes in the middle of
text.
For the Emoji text scrolling benchmark, this reduces
the program changes per frame from > 1000 to around 100.
Use an IFUNC resolver to determine whether we can use
intrinsics for FP16 conversion. This requires the functions
to be no longer inline.
Sadly, it turns out that __builtin_cpu_supports ("f16c")
doesn't compile on the systems where we want it to prevent
us from getting a SIGILL at runtime.
We only have one shader that uses the color2 attribute,
and it doesn't use the uv attribute, so save vertex
memory by putting those in the same space.
This reduce the per vertex space from 32 to 24 bytes.
This reduces the size of our Vertex struct from
48 to 32 bytes. It would be nicer if we could store
the colors in fp16 format in the rendernodes, and
avoid conversion here. But this is still good.
