Previously target device offsets were applied in cairo-surface.c which
could cause bugs when paths were taken as fallbacks, as for example
pointed out by ade55037ff and quick-fixed
by 79190d8985. The quick-fix is now
unnecessary and was removed.
Gah, that was a horrible mistake. It was a flawed hack to create Pixmaps
of the correct depth when cloning patterns for blitting to the xlib
backend. However, it had the nasty side-effect of discarding alpha when
targeting Window surfaces. The correct solution is to simply correct the
Pixmap of the desired depth and render a matching pattern onto the
surface - i.e. a reversal the current acquire -> clone. See the
forthcoming revised xcb backend on how I should have done it originally.
Honour the incoming surface format when we are asked to create a similar
surface with identical content. The goal of
cairo_surface_create_similar() is to create an intermediate with similar
characteristics to the original that can be used in place of the
original and be quick to copy to the original. Matching the format for
the same content, ensures that the blits between the two need only be a
memcpy.
The goal is to create a similar surface with an identical format to
maximise performance in the subsequent blit, e.g. the xlib backend could
make the similar surface with an identical depth and so use the core
protocol, or the image surface could indeed make an identical copy so
that pixman only has to do a fast memcpy. As there is no direct method
to specify such a clone, we ask the backend for a similar surface of
identical content, and trust that the semantics are clear enough for the
intent to obvious.
_cairo_surface_fallback_paint() attempts to avoid a clipped operation if
we can convert the paint into a fill of the clipmask. However by calling
_cairo_surface_fill() we incur a double application of device offset to
the source, triggering various failures.
Company spotted this and managed to extract an excellent minimal test
case, test/clip-device-offset. This commit fixes that failure.
As noted in the comments we could also compute the pattern extents for
gradients with CAIRO_EXTEND_NONE under certain circumstances, i.e.
radial gradients and device axis aligned linear gradients.
XCB avoids the dreaded abort on XError mechanism by forcing the client
to perform deferred error checking. So do so. This allows us to combine
the fire-and-forget rendering model with accurate error checking,
without killing the client or mixing our errors with theirs.
XCB for the win!
Replaying a meta surface can be achieved by using it as a source for a
cairo_paint() so exporting a separate API is unnecesary and confusing.
So after consulting Chris and Carl, we decided to remove the function
again.
Provide default DocumentMedia and PageMedia DSC comments if the user
does not specify them using cairo_ps_surface_dsc_comment(). This is
required so that PostScript viewers such as gv use the correct page
size.
Benjamin Otte pointed out the error of my ways that a clear on a
cairo_image_surface_create_for_data() was not working. This is because I
modified the image surface to skip clears when it knows the target data
has been cleared. This flag must be reset when the user interacts with
the surface, such as providing the initial surface data.
Commit 535bcaa1 moved pthread after win32 to fix mingw build
problems. However this breaks cygwin builds.
On cygwin HAVE_WINDOWS_H is defined but _WIN32 is not. This was
causing windows.h to be included which defines _WIN32. As a result the
win32 code in cairo-misc.c was being compiled but the win32 declaration
in cairint.h was not included.
Fix this by using _WIN32 to enable the win32 mutex functions since
_WIN32 is defined by mingw, visual C++, and winegcc, but not
cygwin. On cygwin, posix functions are preferred as it is a unix
emulation environment.
Gah! I had believed that the dst extents and the clip were correct to
enable unbounded fixup for the unbounded trapezoids. I was wrong, so I
need to requery the trapezoid extents. As this information is already
known, I should update the interface to pass along all relevant
information.
GCC uses a peculiar name for a real 128-bit integer on x86-64.
Speedups, xlib on a gm45
========================
poppler 41246.56 -> 35102.82: 1.18x speedup
swfdec-youtube 12623.01 -> 11936.79: 1.06x speedup
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The autoconf shipping with OpenBSD 4.5 converts sizeof(void*)
to SIZEOF_VOID__ rather than SIZEOF_VOID_P. Work around that
by defining SIZEOF_VOID_P if required.
