A minor thing that I noticed during the testing yesterday, namely that sometimes the SFU subscribed to a video track simulcast layer SimulcastLayerNone as if there was no simulcast on a video track (according to log there were 3 layers). I checked when this could happen. And found that only in 2 cases this is the case:
There are no published simulcast layers (not the case).
If metadata for this track has a width of height equal to 0 (this was a check for audio vs video track).
I replaced the (2) with a more reliable check for the type of media (we actually have always had this information) since it seems to be a better (reliable) way to check if it's a video track or an audio track.
Though we must later investigate if the client can send 0 as width or height for the video track by mistake.
A minor thing that I noticed during the testing yesterday, namely that sometimes the SFU subscribed to a video track simulcast layer
SimulcastLayerNoneas if there was no simulcast on a video track (according to log there were 3 layers). I checked when this could happen. And found that only in 2 cases this is the case:I replaced the (2) with a more reliable check for the type of media (we actually have always had this information) since it seems to be a better (reliable) way to check if it's a video track or an audio track.
Though we must later investigate if the client can send
0aswidthorheightfor the video track by mistake.