Participant Composite #474
Conversation
pkg/gstreamer/bin.go
Outdated
| stateErr <- b.bin.SetState(state) | ||
| }() | ||
| select { | ||
| case <-time.After(stateChangeTimeout): |
There was a problem hiding this comment.
Would a context help here?
There was a problem hiding this comment.
if you're passing the context into other functions/doing more with deadlines I'd agree, but for simple cases like this I prefer the simplicity of time.After. One line less without the defer cancel()
There was a problem hiding this comment.
Indeed. My point would be that I think that ideal (b *Bin) SetState (and potentially callers as well) should take a context as well, allowing the caller to cancel the wait on timeout.
Retrofitting all this is beyond the scope of this PR, of course.
| } | ||
|
|
||
| select { | ||
| case <-time.After(stateChangeTimeout): |
There was a problem hiding this comment.
A context may be a bit more straightforward here as well.
| @@ -326,16 +380,21 @@ func (s *SDKSource) onTrackSubscribed(track *webrtc.TrackRemote, pub *lksdk.Remo | |||
| if !s.initialized.IsBroken() { | |||
| s.mu.Lock() | |||
| switch s.RequestType { | |||
| case types.RequestTypeParticipant: | |||
| s.filenameReplacements["{publisher_identity}"] = s.Identity | |||
There was a problem hiding this comment.
would we make pID a replacement policy too? not sure if that needs to be onTrackSubscribed tho.
There was a problem hiding this comment.
could add ID if/when someone asks, I think for the most part identity is more useful
There was a problem hiding this comment.
yeah you are right.. this is participant composite.. identity is the main thing people would care about.
No description provided.