[Bug 765049] pad: Offset handling inconsistencies

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=765049

--- Comment #14 from Sebastian Dröge (slomo) <slomo at coaxion.net> ---
So the patch seems acceptable as-is now, the funnel problem also was solved as
part of the new patches.


There is just one open question: We now always store the unmodified event
(without pad offset applied). This is the same behaviour as for srcpads but it
has the disadvantage that something like gst_pad_sticky_foreach(sinkpad,
push_to_src_pad) will copy over unmodified events while it should modified
ones. Same goes for gst_pad_get_sticky_event().

I think this boils down to a semantic question. IMHO pad offsets are applied
between pads, so a srcpad should not return the modified event when asked for
it but a sinkpad should. And also pad probes on srcpads and sinkpad should see
the modified event (they do with Vivia's patch).

Main problem with gst_pad_sticky_foreach() modifying the event is... that the
callback is allowed to modify the event. Which means that we would have to
un-apply the pad offset if the callback modified it and then store that back.

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