On Sat, 6 Sep 2014 00:00:10 +0200 Michael Niedermayer <michae...@gmx.at> wrote:
> On Thu, Sep 04, 2014 at 10:40:27PM +0200, wm4 wrote: > > --- > > Follows reimar's suggestion for detecting UTF-16. If the detection goes > > wrong, probing the format won't succeed, so this should not break > > anything. > > > > > I didn't use ffio_ensure_seekback), because this apparently reallocates > > the buffer - which we certainly don't want, because the probe API > > doesn't allow this. > > the format probe API doesnt provide a AVIOContext, and if one is > provided, it could use a allocated and reallocatable buffer > or do i misunderstand ? > > [...] I'm ffio_init_context() to initialize an AVIOContext on a buffer (the probe buffer), so I can use the same code for probing and actual demuxing. And in this case, AVIOContext reallocates the buffer we pass it. This turns the old buffer pointer into a dangling pointer, which breaks probing. Maybe this could be easily fixed by always allocating a new buffer inside the subtitle probe functions, but I'm not sure if this is ok. I'm also not sure why ffio_ensure_seekback() wants to reallocate at all in this case - maybe I should take a closer look. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
