This gets us a fix for not linking with g++ from 1.6.4
We don't take from 1.6.4 the revert of the addition of
missing locking from the GC cache, (meanwhile, we've
already got a corrected fix for this).
Conflicts:
src/cairo-xlib-screen.c
ExtTextOut() does not work with Type 1 font glyph indices when
printing. The same code works fine when dst->dc is a display. It
appears that ExtTextOut expects unicode character values when using
Type 1 fonts and a printer DC.
Fix this by making Type 1 fonts in the win32-printing surface use the
fallback path for non Windows fonts. ie the glyphs will be emitted as
filled paths.
If you think this commit is reminiscent of
40558cb15e, you would be right as it fixes
exactly the same bug I made then and reintroduced in dc714106e1.
So quoting 40558cb:
After consuming the GC we need to unset the clip reset flag, so that
if we try and get a new GC without first putting a fresh one we do not
try to call XSetClipMask on a NULL GC.
This bug first was fixed in 40558cb15e,
but then reintroduced in 9cfd82e87b, which
became part of the 1.6.2 quick release.
As penance to make sure I never repeat this same bug again, I offer this
test case which exercises the XSetClipMask(NULL) path and hopefully
simulates some 'typical' usage of cairo by GUI toolkits.
The beos backend involves a source file written in C++.
Apparently, automake sees the conditional inclusion of
that source file and insists on doing the final link of
cairo with g++ even though that file isn't being compiled
at all.
We definitely don't want that as it makes libcairo link
and depend on libstdc++ unnecessarily, (which can break
distribution packaging when not expecting it, etc.).
This reverts commit 9cfd82e87b.
The fix was broken as it introduced crashes by calling
XSetClipMask with a NULL GC, (basically a reintroduction
of the bug originally fixed here:
7802de6d5edaf998c98b141870dc2c6b4c0f3e91
Let's revert for sake of the release, and fix this correctly
on master.
The Td and Tm operator emulation were setting the font matrix like this:
/some_font [xx yx xy yy x0 y0] selectfont
where [xx yx xy yy] is the font matrix and [x0 y0] is the position of
the first glyph to be drawn. This seemed to be the easiest way to
emulate the Tm operator since the six arguments to Tm required by PDF
are xx,yx,xy,yy,x0,y0.
Before the switch to pdf-operators the font matrix was set like this:
/somefont [xx yx xy yy 0 0] selectfont x0 y0 moveto
The selectfont operator is equivalent to calling findfont, makefont,
and setfont. The makefont operator creates a new font dictionary for
specified font that contains the specified font matrix. The
description of the makefont operator in the PostScript Language
Reference Manual states:
"The interpreter keeps track of font dictionaries recently created
by makefont. Calling makefont multiple times with the same font and
matrix will usually return the same font rather than create a new
one."
So the emulation of Tm and Td was creating a new font dictionary every
time a text string was displayed due to the change in the translation
components of the font matrix. Previously the font dictionary was
re-used as with the translation components of the matrix set to zero,
the font matrix did not change frequently.
Some printers did not handle well the frequent creation a font
dictionary every time a few glyphs were displayed.
Fix this by ensuring the translation components of the font matrix
used in the emulation of Tm and Td operators is always set to
zero. Use moveto instead for the translation components.
(cherry picked from commit c5814d2aa3)
Convert the boilerplate specific flattened content value to the ordinary
CAIRO_CONTENT_COLOR_ALPHA for use with cairo_push_group_with_content() -
otherwise cairo rightfully flags an error and the test harness decides
that the similar surface is not available.
Within the library, we know the precise size of the struct and so can
allocate temporary font options on the stack - eliminating the need
to export an internal alias of cairo_font_options_(create|destory).
We depend on values stored on the context that become invalid upon an
error, so stop processing as soon as an error occurs. Prior to
adjusting, the values returned from the error context, this would cause
an infinite loop whilst calculating the number of segments required for
a tolerance of 0.
Return the default values instead of zero for an error context. This
helps to prevent application logic faults when using the values from
the error context (for an example, see _cairo_arc_in_direction()).
The Td and Tm operator emulation were setting the font matrix like this:
/some_font [xx yx xy yy x0 y0] selectfont
where [xx yx xy yy] is the font matrix and [x0 y0] is the position of
the first glyph to be drawn. This seemed to be the easiest way to
emulate the Tm operator since the six arguments to Tm required by PDF
are xx,yx,xy,yy,x0,y0.
Before the switch to pdf-operators the font matrix was set like this:
/somefont [xx yx xy yy 0 0] selectfont x0 y0 moveto
The selectfont operator is equivalent to calling findfont, makefont,
and setfont. The makefont operator creates a new font dictionary for
specified font that contains the specified font matrix. The
description of the makefont operator in the PostScript Language
Reference Manual states:
"The interpreter keeps track of font dictionaries recently created
by makefont. Calling makefont multiple times with the same font and
matrix will usually return the same font rather than create a new
one."
So the emulation of Tm and Td was creating a new font dictionary every
time a text string was displayed due to the change in the translation
components of the font matrix. Previously the font dictionary was
re-used as with the translation components of the matrix set to zero,
the font matrix did not change frequently.
Some printers did not handle well the frequent creation a font
dictionary every time a few glyphs were displayed.
Fix this by ensuring the translation components of the font matrix
used in the emulation of Tm and Td operators is always set to
zero. Use moveto instead for the translation components.
The experience of running the entire test suite under cairo_push_group()
did at least reveal that there was a potential bug in the pdf backend.
This test is just the simplest of drawing operations - simply filling
the similar surface with solid blue, and then blitting that surface to
the destination.
Cut'n'paste from commit c1f765:
Without these checks, a user could hit an assertion failure by passing
a finished surface to cairo_surface_write_to_png_stream. Now we return
a nice CAIRO_STATUS_SURFACE_FINISHED error in that case instead.
It is possible for _cairo_hull_prev_valid to be called just once
right before the calling loop is going to terminate. In this
case we really don't want to walk off the beginning of the
array and start wandering.
Thanks to Jonathan Watt for noticing this problem:
https://bugzilla.mozilla.org/show_bug.cgi?id=306649#c21
This reverts commit 07122e64fa.
The extra testing did find a pdf bug, and that should be fixed,
but the extra maintenance burden of running another iteration
of all tests does not seem justfied at all---particularly since
it looks like dozens of new reference images would be needed
for the svg backend.
Also, the new "failures" of the image backend with this new
testing look like a misunderstanding of exactly what the new
testing is actually drawing.
Without these checks, a user could hit an assertion failure
by passing a finished surface to cairo_surface_write_to_png.
Now we return a nice CAIRO_STATUS_SURFACE_FINISHED error in
that case instead.
As identified by Vladimir Vukicevic,
_cairo_xlib_surface_create_similar_with_format() was erroneously passing
down the source Visual when creating a surface with a different
XRenderPictFormat.
If the xserver doesn't support the RENDER extension or simply doesn't
have the matching PictFormat then xrender_format might be NULL. Check
and fallback in this case.
Test surfaces using similar surfaces with both CONTENT_COLOR and
CONTENT_COLOR_ALPHA, if applicable. This seems justified by the apparent
bugs in the pdf backend when going from an ARGB32 similar surface to
a destination RGB24 surface as well as isolated bugs in the image
backend.
The original goal was to try and trick the test suite into producing
a xlib surface with mismatching Visual/XRenderPictFormat. This succeeds,
although with a little bit of brute force in the xlib backend, but the
search to reproduce a BadMatch error fruitless.
