[FFmpeg-cvslog] r15583 - trunk/libavformat/dv.c

Mike Melanson mike
Wed Oct 8 06:04:11 CEST 2008


romansh wrote:
> Author: romansh
> Date: Tue Oct  7 18:14:33 2008
> New Revision: 15583
> 
> Log:
> simplifying code as per Michael's suggestion

One of the revisions between 15580 and 15583 broke compilation of gcc 
2.95.3 on x86_32. I've omitted the irrelevant warnings here:


/usr/local/gcc-2.95.3/bin/gcc -DHAVE_AV_CONFIG_H -D_FILE_OFFSET_BITS=64 
-D_LARGEFILE_SOURCE -I. -I"/home/melanson/fate/source" -D_ISOC99_SOURCE 
-D_POSIX_C_SOURCE=200112 -fasm -fomit-frame-pointer -Wall -Wno-switch 
-Wpointer-arith -Wredundant-decls -Wcast-qual -Wwrite-strings -O3 
    -c -o libavformat/dv.o /home/melanson/fate/source/libavformat/dv.c
In file included from /home/melanson/fate/source/libavformat/dv.c: In 
function `dv_extract_audio_info':
/home/melanson/fate/source/libavformat/dv.c:204: warning: assignment 
makes integer from pointer without a cast
/home/melanson/fate/source/libavformat/dv.c:204: parse error before `['
/home/melanson/fate/source/libavformat/dv.c:190: warning: `ach' might be 
used uninitialized in this function
make: *** [libavformat/dv.o] Error 1

So it looks like this revision is the problem, and here's the problem line:

> +    ach = (int[4]){  1,  0,  2,  4}[stype];

BTW, before anyone goes all language-lawyer and wants to argue about 
whether this is truly and philosophically valid C, I would like to 
remind you that 2.95.3 is still officially supported by FFmpeg. Ergo, C 
code must compile under 2.95.3. That's the law of FFmpeg.

-- 
     -Mike Melanson




More information about the ffmpeg-cvslog mailing list