sticky events

[Stefan Sauer]

On 03/02/2014 09:44 PM, Stefan Sauer wrote:
> hi,
>
> I tried to do the proper pad blocking thing in buzztrax again, now with
> 1.0. One issue I face right now when adding sources (or sources +
> transforms) to adder/audiomixer is that the new elements are not
> negotiated. While debugging I was wondering whether sticky events are
> replayed on the new request (sink)-pad of adder. They are not. In tee
> this is done. If I simply create the counterpart of the tee code in
> adder (https://gist.github.com/ensonic/b73c16958d9304dfd2e0), I get
> (lt-buzztrax-edit:6264): GStreamer-WARNING **: pad '':sink_2 pushing
> stream-start event in wrong direction
> (lt-buzztrax-edit:6264): GStreamer-WARNING **: pad '':sink_2 pushing
> caps event in wrong direction
> (lt-buzztrax-edit:6264): GStreamer-WARNING **: pad '':sink_2 pushing
> segment event in wrong direction
>
> Now I am a bit confused:
> 1.) how come we store these events on the sticky list of the src-pad?
> How would one use gst_pad_push_event() to sent them there in the first
> place?
Now after some chat on irc, I learned that we actually don't have any
upstream stick events right now. Still these are stored on the pad, a bug?

Stefan
> 2.) Why does work for tee and not for adder?
> 3.) How does it even make push events on newly created request pads as
> they don't have peers yet.
>
> If someone can shed some light here, I can apply this to the docs for 
> gst_pad_sticky_events_foreach() which miss the obvious part that
> explains "when you wnat to use this function" :/
>
> Stefan
> _______________________________________________
> gstreamer-devel mailing list
> gstreamer-devel at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel