[FFmpeg-devel] [PATCH 1/3] Add av_build_index

Ronald S. Bultje rsbultje
Mon Aug 16 16:08:36 CEST 2010

Hi Michael,

On Sun, Aug 15, 2010 at 11:53 PM, Michael Chinen <mchinen at gmail.com> wrote:
> On Mon, Aug 16, 2010 at 3:10 AM, Ronald S. Bultje <rsbultje at gmail.com> wrote:
>> Hi,
>> On Sun, Aug 15, 2010 at 2:43 AM, Michael Chinen <mchinen at gmail.com> wrote:
>>> Okay, this updated patch plus this 0004 patch for asfdec.c which
>>> depends on 0001 (but is independent of 0002 and 0003) will give
>>> support for asfdec.c and allow formats without parsers to work by
>>> introducing a new virtual function into AVInputFormat for flushing.
>> (I might be wrong here, but I always thought the seek virtual function
>> was supposed to take care of this for some formats. That's indeed
>> confusing, since for others it does the actual seeking itself, but
>> making that more messy isn't quite the right thing, so you might have
>> to separate that in a more generic way here, i.e. updating all
>> seek-as-flush vfuncs to be a flush function.)
> For some formats (like asf,) the read_seek function pointer does a
> flush and a seek, but it also does some index building.
> But if I understand you correctly, some formats' read_seek do flushing
> without seeking? ?I'll be looking, but does anyone know an example I
> could look at? ?I wonder if there is some flag or other way of
> catching all of these. ?I agree that moving these functions to
> read_flush would clean things up.

As said, I'm not even sure if this is really the case, so if nobody
replies, assume I made it up and just go on as before.


More information about the ffmpeg-devel mailing list