Fix Rtp padding generator for older firefox build. #1152
Conversation
Also set marker to false per RFC
|
does these changes still works for Chrome? I think to remember we had some issues in the past with the marker set to true (but I might be wrong). |
| @@ -100,6 +100,7 @@ void RtpPaddingGeneratorHandler::sendPaddingPacket(std::shared_ptr<DataPacket> p | |||
| RtpHeader *rtp_header = reinterpret_cast<RtpHeader*>(padding_packet->data); | |||
|
|
|||
| rtp_header->setSeqNumber(sequence_number.output); | |||
| rtp_header->setMarker(false); | |||
There was a problem hiding this comment.
could we set the marker bit inside RtpUtils::makePaddingPacket please?
There was a problem hiding this comment.
I'm sorry idk why i messed up here
| sendPaddingPacket(packet, last_padding_packet_size_); | ||
| // Temporary fix since https://bugzilla.mozilla.org/show_bug.cgi?id=1435025 is fixed | ||
| // Only send full rtp padding packet as suggested also in webrtc code. | ||
| // sendPaddingPacket(packet, last_padding_packet_size_); |
There was a problem hiding this comment.
this could affect the way the congestion control works, have you tested it with Chrome too? did you notice any significant change in the amount of time it takes for simulcast to increase the quality layers?
|
I tested with chrome and seems ok! also because chrome itself send the packet with the marker false. Also, I don't think that the minimum padding size that we drop here will affect much the increase rate of bw estimation (but I could be wrong). If you can also test this will be great, but I don't think this will cause any issue. |
Also set marker to false per RFC
Idk if it needs some unit tests
[] It needs and includes Unit Tests
[] It includes documentation for these changes in
/doc.