Optimizing compilers aggressively remove code that is executed only
after an undefined behaviour occurred.
Also, the difference of two (non char) pointers hides an integer
division that, because the divisor is known at compile time, is
transformed into a multiplication by a pseudo-reciprocal, and in this
case the difference is not always a multiple of the divisor, resulting
in an invalid comparison predicate.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=74779
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
In Hobby's paper it is proved that INTERSECTION events can be
processed in any order by ignoring intersections between edges
non-adjacent in the active edges list.
But with respect to START/STOP events they must be processed in
order. Because START/STOP events have always exact y, it is
sufficient to know whether an integer y intersection is a
default/excess approximation of the exact to properly sort events.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=74779
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Quells the following warning:
cairo-script-scanner.c: In function ‘_translate_string’:
cairo-script-scanner.c:1623:18: warning: comparison between signed and
unsigned integer expressions [-Wsign-compare]
if (buf_len <= 8 + 2*string->len) {
^
Even features which are disabled by default should appear in
cairo-features.h.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=83359
(Patch approved by Behdad Esfahbod in bug tracker.)
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Recent versions of GCC/clang feature a new set of compiler intrinsics
for performing atomic operations, motivated by the operations needed to
support the C++11 memory model. These intrinsics are more flexible than
the old __sync_* intrinstics and offer efficient support for atomic load
and store operations.
Having the load appear atomic to the compiler is particular important
for tools like ThreadSanitizer so they don't report false positives on
memory operations that we intend to be atomic.
Patch from Nathan Froyd <froydnj@mozilla.com>
To maintain compatibility with OpenGL ES 2.0, the matrix in
_cairo_gl_shader_bind_matrix() should be manually transposed,
and GL_FALSE passed as the transpose argument to the
glUniformMatrix3fv() call as it is the only valid value for
that parameter in OpenGL ES 2.0.
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Acked-by: "Henry (Yu) Song" <henry.song@samsung.com>
An earlier fix to bug 84324 added a regression in the font size of
canvas text in Inkscape when compiled with the Quartz backend.
Patch from Andrea Canciani
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=84324
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
A common source of error when people are setting up pango on a new
device is when font files don't have the right permissions and cannot
be opened. Cairo was returning out-of-memory before, making the
Pango error message useless.
With this change, cairo will return file-not-found, and pango prints
that out. It's still not ideal; a ENOACCESS equivalent would have
been better.
Otherwise, calling cairo_set_user_data(cr, key, 0, 0) many times
causes a long user data array, almost all of whose slots are empty.
It leads to unnecessarily much memory consumption and long execution time of
cairo_set_user_data(cr, key, 0, 0) and cairo_get_user_data(cr, key) after
it.
This issue probably happens since the commit
http://cgit.freedesktop.org/cairo/commit/?id=9341c254a
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Commit 4e3ef57bc8 added
coverage-intersecting-triangles with an incorrect reference and
generator. The test checks the rasterization of two overlapping
triangles in the following position:
. .
|\ /|
| X |
|/ \|
.---.
Since the triangles have both vertical and horizontal sides of size
x/WIDTH, the expected coverage is 3/4 (75%) of (x/WIDTH)^2. The
original code, instead, was checking for a coverage of 0.75*x/WIDTH,
as if one of the sides was always 1 unit long.
The image and xlib backends still suffer from some jitter, caused by
the approximation of the actual coverage by means of sampling. For
this reason their references are still considered XFAIL, even though
their result now looks mostly consistent with the expected reference.
The make-cairo-test-constructors.sh script executes several commands
without checking their success. This can lead to undetected errors,
like those fixed in 86fad78fcd.
The script now exits with an error status if no file is
input. Moreover, it sets the '-e' flag, so that if a command fails,
the whole script is immediately terminated with an error.
In the Makefile.am, the script result is now checked and the target
file is removed upon error. This ensures that the
'cairo-test-constructors.c' target completes succesfully only if no
error occurred.
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
Since 70cc8f250b the quartz backend is
using some APIs that are not available on MacOSX 10.4 directly
(i.e. without detecting their availability through dynamic linking).
This means that the quartz backend does not work anymore on MacOSX
10.4 and that the 10.5 SDK (or newer) is needed to build.
files. Because of this, the generated cairo-test-constructor only
contained a few tests and the test suite was thus incomplete.
Original patch by Andrea Canciani <ranma42@gmail.com>
The bug mentioned in KNOWN_ISSUES was fixed in
2b3d8de11a, but the documentation was
not updated. Moreover, the header of KNOWN_ISSUES was 1.10-specific.
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
On AIX 6.1, getdelim() and getline() are not provided by default,
causing a gcc compilation error. With _GETDELIM defined, AIX's stdio.h
header provides definitions for these routines.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=89356
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
CGFontGetGlyphPath was not public and is not available anymore on
modern OSX/iOS systems. The same functionality is available through
the CoreText API since OSX 10.5.
Based on a patch by Simon Cozens.
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=84324
Beginning in 1.12, a crash can occur when the win32 surface's image size
does not cover the entire surface (e.g. due to clipping regions).
This patch enlarges the fallback surface created of the amount
necessary not to write past the end of the DIB. It assumes that
Clip applied to an HDC are clamped to (0,0,width,height) of the HDC.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=53121
Reviewed-by: Bryce Harrington <bryce@osg.samsung.com>
The current code results in trace lines with the source surface being
pushed on the stack two times instead of one:
s1 s1 //ARGB32 48 48 similar-image % s2
Instead of:
s1 //ARGB32 48 48 similar-image % s2
This greatly confuses later commands when the script was replayed,
causing traces for trivial GTK3 programs to be unplayable, usually
yielding the following error:
"invalid value (typically too big) for the size of the input (surface, pattern, etc.)"
Drop the duplicated entry from the trace line printed by the
cairo_surface_create_similar_image() override.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=73580
Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>
