[Ffmpeg-devel] [RFC] [PATCH] FLC/FLX/DTA encoder

[Alex Beregszaszi]

Hi,

> theres trailing whitespace and tabs in the patch ...

Yep and hard tabs. Will be removed of course.

> > +static int palette_sort(const void *a, const void *b) {
> > +    int ia = *(int*)a, ib = *(int*)b;
> > +    
> > +    if (ia < ib) return -1;
> > +    else if (ia == ib) return 0;
> > +    return 1;
> 
> doesnt 
> return *(int*)a - *(int*)b;
> work? (yes it can overflow so i dunno ...)

As you pointed later, this code wont be used anymore.

> palettes in AVFrame..data[1] are stored in native endian, i would definitly
> be in favor of unifying the format with mplayer but thats a seperate issue

Ah, yes. Since I sold my iBook I could test it on BE, so possibly it has
some endianess issues.

> something like:
> 
> j=0;
> for (i = 0; i < 256; i++)
>     if (pal[i] != oldpal[i])
>         changetable[j++] = i+1; // slot 0 may change too
> 
> looks nicer ...

lol, of course

> iam protesting against this missuse of extradata
> its the muxers job to write the header from width/height/... not to missuse
> extradata, and yes the demuxer must be fixed too, if there are no volunteers
> ill fix the demuxer and decoder
> from a quick look i dont see why fl* would need to touch extradata at all
> width/height/codec_tag/bits_per_sample seem enough

Just have a look at the decoder part. The decoder determines some hacks
based on the extradata (some chunks in the header). Anyway, it would be
handy fixing this issue.

--
Alex Beregszaszi