Philipp Hancke

Has anyone filed an issue with brave? I understand the intention to limit the ip gathering but this solution is quite... harmful

at first I thought this was a chrome bug as [the spec](http://rtcweb-wg.github.io/jsep/#rfc.section.3.5.4) is pretty clear that the candidates should only be surfaced via the candidate event. As looking at the...

> Most routers have timeouts on address mappings, which means the connection information in the server-reflexive ICE candidates expires. WebRTC implementations should periodically ping the STUN server to keep the...

actually it does, this is called "m-line recycling" (and there have been spec changes recently). Implementations may vary though.

This is something the browser would do when you call transceiver.stop() which marks the m-line in the SDP as inactive. - which isn't implemented in Chrome yet. If the problem...

Chrome does limit mid to 32 bytes (wanted 16 but... this broke someone) [since M98](https://bugs.chromium.org/p/webrtc/issues/detail?id=12517). Not sure if this also applies to rid, tagging @alvestrand see also https://mailarchive.ietf.org/arch/msg/mmusic/_Kew1cU1-xK6mGfVebMi1-00_EM/

see also #1979

isn't that related to chrome's continued lack of compliance in its so-called "spec-compliant simulcast"?

see #2155

I think https://jsfiddle.net/fippo/s1fguya7/3/ reproduces this. The current direction should be sendOnly and negotiationneeded should fire, suggesting you need to do another negotiation.