On 05/25/11 12:29 pm, Diego Biurrun wrote:
On Thu, May 12, 2011 at 07:02:01PM -0700, Dave Yeo wrote:
On 05/12/11 08:30 am, Diego Biurrun wrote:

Now on to specific FATE configurations and their flags and hacks:
[...]
* os/2

This configuration sets '-D_GNU_SOURCE', which is suspicious.
I'd like to see what happens now if it is removed.
[...]
CC      libavcodec/aaccoder.o
/usr/src/ffmpeg.libav/libavcodec/aaccoder.c: In function 'coef2minsf':
/usr/src/ffmpeg.libav/libavcodec/aaccoder.c:522:5: error: implicit
declaration of function 'log2f'
[...]
make: *** [libavcodec/aaccoder.o] Error 1

This really should be caught in configure where check_mathfunc log2 and
log2f generate implicit declaration of function warnings. One fix would
be enabling -Werror=implicit-function-declaration earlier in configure.

That fix you suggest is less trivial than it may sound.

What about the attached? OS/2 (actually eCS) is now shipping with GCC 4.5.2 which is plenty new enough to support -Werror=implicit-function-declaration and it can be removed if and when the headers are fixed.
I've filed http://svn.netlabs.org/libc/ticket/243 about this.

Can you paste config.log and config.h somewhere?  And maybe upload a
tarball of system headers?

Sure, they should be here, http://www.box.net/shared/rm86u32ey0. Note that there are a couple of headers that I've added such as pthread.h. The headers are also available at http://svn.netlabs.org/libc/browser/branches/libc-0.6/src/emx/include.

 What libc does OS/2 use?

It's a fork of the ancient EMX libc (http://en.wikipedia.org/wiki/EMX_%28programming_environment%29), http://svn.netlabs.org/libc/wiki.
Dave
diff --git a/configure b/configure
index c628ff6..ede1695 100755
--- a/configure
+++ b/configure
@@ -2477,6 +2477,7 @@ case $target_os in
     os/2*)
         ln_s="cp -f"
         objformat="aout"
+        add_cflags -Werror=implicit-function-declaration
         add_ldflags -Zomf -Zbin-files -Zargs-wild -Zmap
         SHFLAGS='$(SUBDIR)$(NAME).def -Zdll -Zomf'
         FFSERVERLDFLAGS=""
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to