Closed o-reo closed 5 years ago
I think that the problem as you said, is caused by the infinite loop rather than latency not being calculated. For example if one of the sources is blocked or has an elevated timeout when receiving frames, all the pipeline will be blocked like in #18.
A solution could be exit the loop after some time, and fixate a default caps. But I'm not sure it will not fail again when fixating the caps or when not receiving frames in the create function.
Adding a timeout parameter and sending gap events could do the trick. I will do some testing.
I see it's possible to use NDI in pull mode with framesync. It would probably solve our problem.
This should be fixed by https://github.com/teltek/gst-plugin-ndi/pull/34
@rubenrua this can be closed
In the case where the NDI receiver does not return the expected frame type or does not get any frame the latency query fails.
example pipeline, note that the NDI stream only sends video frame :
As the latency is not calculated, the muxer does not process the video frames as it waits indefinitely for the audio frames.
when setting debug env with export
GST_DEBUG=flvmux:6,aggregator:6
you get the following line:I assume setting a default latency other than none should solve the problem ? It may also be due to the infinite loop in fixate functions.