No point in passing in a column of zeroes that we multiply by zero, or
the row that we drop on the floor. We could do even better by just
passing in a 3x2, at the expense of adding a language version
dependency (or just pass in 2 vectors).
No significant performance difference.
The shift primitive of the Solaris 9 Bourne shell doesn't like
it if we try to shift more formal arguments than are there.
This causes the compare_versions function in autogen.sh to croak when
the actual and compared versions differ in the number of
components.
We're not supposed to be redefining PACKAGE_VERSION, PACKAGE_...
from the configure generated confdefs.h. This patch rudely adds
paper over the problem. The compiler warnings are a problem for
us since our checking of various compiler flags assumes that
no news is good news, and that any warning messages are due
to the flags under test. The regression appears when using
an autoconf >= 2.64, at least, but not with 2.61.
The same issue appears in the pthread test because our conftest
unconditionally #defines _GNU_SOURCE, but autoconf ends up doing
that in the confdefs.h.
Use two levels of pthread support: a minimal level used to
build cairo itself, and a full level to build threaded apps
which want to use cairo. The minimal level tries to use
pthread stubs from libc if possible, but falls back to the
full level if that's not possible. We use CFLAGS=-D_REENTRANT
LIBS=-lpthread to find a real pthread library since that seems
to work on every unix-like test box we can get our hands on.
Introduce a new CAIRO_CC_TRY_LINK_WITH_ENV_SILENT macro for running
generic link tests with arbitrary CFLAGS/LIBS/LDFLAGS and no muttering
of autoconf messages. Rewrite the previous CAIRO_CC_TRY_FLAG in terms
of it.
What we want to use is size_t, but we don't want the implied POSIX
dependency. However, POSIX does say that size_t is an unsigned integer
that is no longer than a long, so it would appear safe to use an
unsigned long as a replacement. Safer at least than unsigned int.
Currently wether internal glew is built is dependant on wether the
tarball in use was built with internal glew or not. And that's not how
it should be.
This is an attempt to fix the broken situation we've been in where
automake links libcairo.la with c++ because it might potentially maybe
include C++ files.
Those potential files only exist in Chris' throwaway backends (skia, qt)
and the BeOS backend, so for 99.99% of cases, these backends are not
needed and linking with c++ is overkill. Also, no one wants to have
libcairo.so link to libstdc++.
This patch fixes that in mutliple steps:
1) Add build infrastructure to distinguish between C and C++ backends.
This is done by allowing to specify backend_sources as well as
backend_cxx_sources variables in Makefile.sources.
2) Optionally build a libcairo_cxx.la noinst library
This intermediate library is built for C++ backends only and therefor
linked using c++. It is then linked into the final libcairo.la. This
does not require c++, so the linking of libcairo.la is done with cc.
This also works around various weirdnesses that the current build system
exposes, where it assumes cisms when in fact using c++ semantics, like
not detecting c++ properly or:
https://bugzilla.redhat.com/show_bug.cgi?id=606523
Add general docs and document acquire/release. I'm not happy with the
documentation yet. In particular, I have 2 issues:
1) The threading guarantees Cairo provides are missing.
2) There's no docs on which Cairo functions do acquire devices.
I guess I'll have to fix them later.
I did this manually so I could review the docs at the same time.
If anyone finds typos or other mistakes I did, please complain to me (or
better: fix them).
Use cairo_drm_xr_sources/headers for the drm_xr stuff. This makes it
work correctly in gtk-doc instead of conditionally adding it to
cairo_drm_sources/headers.
