Fwd: [gst-devel] Proposition for a media test suite

[Edward Hervey]

forgot to reply to all :(

---------- Forwarded message ----------
From: Edward Hervey <bilboed at gmail.com>
Date: Nov 14, 2005 5:54 PM
Subject: Re: [gst-devel] Proposition for a media test suite
To: "Ronald S. Bultje" <rbultje at ronald.bitfreak.net>


Hi,

On 11/14/05, Ronald S. Bultje <rbultje at ronald.bitfreak.net> wrote:
> On Mon, 2005-11-14 at 17:30 +0100, Edward Hervey wrote:
> > On 11/14/05, Ronald S. Bultje <rbultje at ronald.bitfreak.net> wrote:
> > > Read the question again, you didn't answer the main part. How do I make
> > > sure the first frame coming out of my pipeline is the one I want to seek
> > > to when seeking in a decodebin-generated pipeline?
> >
> >   By comparing the seek start time and the timestamp of the first
> > buffer that arrives in the sinks after doing a GST_SEEK_FLAG_ACCURATE
> > | GST_SEEK_FLAG_FLUSH seek when the pipeline is completely set ? Sorry
> > for not detailling more previously, I thought that part was obvious.
>
> You still didn't read my question, I explained the problem above. There
> is *no pipeline to seek on* since decodebin only builds it when it goes
> to _PAUSED and I want to seek before going to _PAUSED (e.g. in _READY,
> as you suggested).

Quote from my initial response :
""  Seek in STATE_READY : the seek event is taken into account for when
switching to PAUSED""

  Decodebin should remember that event, and when constructing it's
internal pipeline, as soon as it has the first element, it should send
it to that element (gst_element_send_event()) when that element is in
READY.

  Demuxers and parsers should be able to receive seek events when in
READY, in order to modify then initial playback range (which is the
whole stream in non-segment mode).

  This time I'm not being sarcastic, It wasn't clear at all, thanks
for spotting this out. And indeed I need that behaviour for editing.
So I guess me or Wim will be adding that anytime soon to decodebin and
the demuxers.

  Edward


>
> I'm asking for better-defined behaviour because we've always had those
> seek flags and we all know what they mean (it's not like gtk-doc
> magically makes the world go around, y'know? It just documents that it
> does); and yet it never worked...
>
> Cheers,
> Ronald
>
>


--
Edward Hervey
Junior developer / Fluendo S.L.
http://www.pitivi.org/


--
Edward Hervey
Junior developer / Fluendo S.L.
http://www.pitivi.org/