[FFmpeg-cvslog] r19234 - trunk/libavcodec/wma.c

Sascha Sommer saschasommer
Sun Jun 21 14:25:04 CEST 2009


Hi,

On Samstag, 20. Juni 2009, Ivan Kalvachev wrote:
> On 6/20/09, Sascha Sommer <saschasommer at freenet.de> wrote:
> > Hi,
> >
> > On Samstag, 20. Juni 2009, Ivan Kalvachev wrote:
> >> On 6/20/09, faust3 <subversion at mplayerhq.hu> wrote:
> >> > Author: faust3
> >> > Date: Sat Jun 20 13:06:48 2009
> >> > New Revision: 19234
> >> >
> >> > Log:
> >> > Simplify run level decoding:
> >> > - remove unneeded vlc code < 0 check
> >>
> >> Hum, would you explain this a little bit more.
> >> Afaik -1 is returned when sequence that doesn't belong
> >> to any known vlc is met.
> >> This usually happens on broken streams.
> >>
> >> Can you prove that wma vlc don't have invalid sequences?
> >
> > In libavcodec/bitstream.c build_table() the table for get_vlc2 is built
> > with unused codes set to -1
> >
> >     for(i=0;i<table_size;i++) {
> >         table[i][1] = 0; //bits
> >         table[i][0] = -1; //codes
> >     }
> >
> > After the tables are filled none of the table entries are -1 anymore.
>
> OK, if you have checked all tables.
> I still would like at least a comment in the code explaining this.
>

I'm not the maintainer of wma.c but I wouldn't mind if you add one.

> BTW, why all code comments are in doxygen style? /** */?

I think comments in new patches have to be doxygen compatible. As my changes 
modified most of the function, I decided to make the comments doxygen 
compatible to save a bit of work in case someone wants to go over wma.c and 
fix the things that are usually brought up when submitting new patches.

Regards

Sascha



More information about the ffmpeg-cvslog mailing list