Now that I have finally fixed the BSD CPPFLAGS issues in a clean way
I believe it is time to review our FATE configurations again and
possibly unify them.

First off, it would be great to have a Cygwin FATE configuration again.
People do use Cygwin and we would profit from timely notification of
breakage.

Some FATE configurations enable postproc, others don't.  I just
switched the defaults so that postproc gets enabled automatically
if --enable-gpl is passed to configure.  So, FATE maintainers, please
drop --enable-postproc from your configure lines.

There are some FATE configurations that use --disable-debug and some
that use --enable-shared - why?

Reinhard just tells me that he would like to see FATE boxes run make in
verbose mode, i.e. with 'V=1' set so that all compiler flags are visible
in the build logs.


Now on to specific FATE configurations and their flags and hacks:

* freebsd and dragonfly bsd:

The '-D__BSD_VISIBLE -D__XSI_VISIBLE=600' flags should no longer be
necessary.  I have tested compilation on a freebsd box myself, so
I'm confident that this will really work.  Please remove.

* freedos

With a bit of luck '-U__STRICT_ANSI__' can be dropped now, please test.

* os/2

This configuration sets '-D_GNU_SOURCE', which is suspicious.
I'd like to see what happens now if it is removed.

* avr32 / linux

This configuration uses --disable-demuxer=yop, probably to work around
a compiler bug.  Is this still necessary?

* mips64 / linux

This configuration uses --extra-cflags=-mno-fused-madd, probably to work
around some bug.  Is this still necessary?

* sparc64 / linux

Here configure is fed --disable-vis, which makes me think we are missing
something in configure that should be done instead.

* sparc64 / openbsd

This configuration adds --disable-pthreads to the configure line.  Why?

* sparc64 + x86_32 / sunos

Here --extra-libs=/usr/lib/64/values-xpg6.o gets added.  I wonder if this
is still necessary and/or if we can do it in configure.

* x86_32 + x86_64 / sunos / gcc 4.6.0 + gcc 4.7.0

On this platform we see --disable-vdpau added to configure.  What is going
wrong here with vdpau?  Can it be solved by upgrading vdpau or do we need
more elaborate configure tests?

* x86_32 + x86_64 / linux / suncc

This configuration adds empty host CFLAGS with '--host-cflags=' and passes
'--disable-asm' to configure.  Why?  Is there something we should have
more or better tests for in configure?

* x86_32 + x86_64 / mingw32

The configure line contains --enable-w32threads here, but I wonder if
pthreads cannot be used on Windows?  I thought the answer was yes...

* x86_32 / mingw32

Here we have --enable-memalign-hack being added to the configure line.
Why?

* x86_32 / linux / valgrind   vs.   ppc / linux / gcc-valgrind

The latter passes --workaround-gcc296-bugs=yes to valgrind, while the
former does not.  This is weird.


OK, that's it for now.  More suggestions welcome...

Diego
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to