[Bug 797162] baseparse: Segment events are not necessarily a discontinuity

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=797162

--- Comment #4 from Jan Schmidt <thaytan at noraisin.net> ---
(In reply to Sebastian Dröge (slomo) from comment #3)
> (In reply to Nicolas Dufresne (ndufresne) from comment #2)
> > Doesn't baseparse uses an adapter ? GstAdapter does not have what's needed
> > to properly track time when buffers from multiple segments are in the
> > adapter. Do we need to drain in order to place the new segment at output at
> > the right place.

The main problem is that I can't think of the use case for the existing
behaviour, and there's no test case for it. I suspect this code should have
been rewritten when segments were.

> That indeed needs some checking, but then also needs some detection to see
> if the segment actually changes anything with regard to sync so that this is
> needed at all.

If someone can define what baseparse and adapter should do exactly, then we
could write some unit tests and implement it.

> > The discont flag though is not needed.
> 
> What do you mean? It's what is signalling that this buffer starts
> "elsewhere" (independent from time information though, there's another
> independent flag for that).

He was agreeing with me that the output from base parse should have a discont
flag only if the input did. baseparse shouldn't add it just because it got a
segment event.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.