theora dev - Feb 2005 - [patch] fixes for avi2vp3

here is a patch that fixes various bugs/problems with avi2vp3:

1. open() is not called with O_BINARY , this causes the program to
crash or give undesired results on win32 (and maybe other
platforms?), as binary mode is not default.

2. win32 does not understand chmod flags, ie:
(S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH)

3. buffer = malloc(AVI_max_video_chunk(avifile));
for some reason, AVI_max_video_chunk(avifile) returns 0, i've
set an arbitrary high value for it until someone fixes it
correctly.

-- Nehal
-------------- next part --------------
A non-text attachment was scrubbed...
Name: avi2vp3-fixes-nehal.diff
Type: application/octet-stream
Size: 1856 bytes
Desc: not available
Url :
http://lists.xiph.org/pipermail/theora-dev/attachments/20050226/2e1f7cef/avi2vp3-fixes-nehal.obj

On Sat, Feb 26, 2005 at 07:45:35PM -0800, Nehal wrote:
> here is a patch that fixes various bugs/problems with avi2vp3:
Cool. Thanks. I've committed a variation on your patch. Can you let me 
know if it works?
> 2. win32 does not understand chmod flags, ie:
> (S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH)
Google suggested one can still pass a literal mode flag (0644) so I did 
this instead?
> 3. buffer = malloc(AVI_max_video_chunk(avifile));
> for some reason, AVI_max_video_chunk(avifile) returns 0, i've
> set an arbitrary high value for it until someone fixes it
> correctly.
Right. It looks like it returns zero for files without an idx chunk, so 
one only needs to fallback if it does return zero. This should be more 
robust than the generic 32K buffer that went into the theora-tools 
version.

 -r