in H264 and H264, parameters (VPS, SPS and PPS) are automatically
prepended to random access units. When parameters were nil, they were
prepended anyway, leading to a crash. Parameters are not prepended only
if they are valid.
* add webrtcp static tcp mux port
* add ice nat1 host configuration and cleanup
* typo
* rename config keys
* apply codecov suggestions
* apply review suggestions
* typo
* dont use deepequal for WebRTCICETCPMuxAddress
* unexport NewPeerConnection()
* remove Dockerfile
* use an empty list instead of nil value in webrtcICEHostNAT1To1IPs
* drop webrtcICETCPMuxEnable and enable TCP mux when webrtcICETCPMuxAddress is filled
* run go mod tidy
Co-authored-by: aler9 <46489434+aler9@users.noreply.github.com>
When latency is high, one side of the peer connection switched to the
"connected" state before the other one, and then closed the WebSocket
connection since it's useless after the peer connection has been
established. This caused the other side of the connection to detect a
WebSocket error and to exit.
The WebSocket connection must remain open, otherwise the
"connected" state is not set by both parts.
Fixes#1103
gortsplib/v2 supports multiple formats inside a single track (media). This allows to apply the resizing algorithm to single formats inside medias.
For instance, if a media contains a a proprietary format and an H264 format, and the latter has oversized packets, they can now be resized.
* API: number of bytes received/sent from/to RTSP connections
* API: number of bytes received/sent from/to RTSP sessions
* API: number of bytes received/sent from/to RTMP connections
* API: number of bytes sent to HLS connections
* API: number of bytes received from paths
* metrics of all the above