Question about TRANSPORT_PRODUCE.

mediasoup libraries
1

From the code from transport.cpp:

	const auto& producerRtpHeaderExtensionIds = producer->GetRtpHeaderExtensionIds();

				if (producerRtpHeaderExtensionIds.mid != 0u)
				{
					this->recvRtpHeaderExtensionIds.mid = producerRtpHeaderExtensionIds.mid;
				}

				if (producerRtpHeaderExtensionIds.rid != 0u)
				{
					this->recvRtpHeaderExtensionIds.rid = producerRtpHeaderExtensionIds.rid;
				}

				if (producerRtpHeaderExtensionIds.rrid != 0u)
				{
					this->recvRtpHeaderExtensionIds.rrid = producerRtpHeaderExtensionIds.rrid;
				}

				if (producerRtpHeaderExtensionIds.absSendTime != 0u)
				{
					this->recvRtpHeaderExtensionIds.absSendTime = producerRtpHeaderExtensionIds.absSendTime;
				}

				if (producerRtpHeaderExtensionIds.transportWideCc01 != 0u)
				{
					this->recvRtpHeaderExtensionIds.transportWideCc01 =
					  producerRtpHeaderExtensionIds.transportWideCc01;
				}

It seems if a transport has more than one producer and if the producers have different headerExtensionIds the new one will overlap the old ones?

Lets say producer 1

a=extmap:4 urn:ietf:params:rtp-hdrext:sdes:mid

and producer 2

a=extmap:8 urn:ietf:params:rtp-hdrext:sdes:mid

Do I miss something here or its a bug?

2

Wellcome to SDP hell.

Yes, in SDP design.

2 Likes
3

Thanks for your reply, would you mind give me some references or what would happend if i pass two diffrent headerExtensionID to mediasoup?

4

Unexpected things (such an issues in previous Producers/Consumers) since that should not happen. New ext ids will replace previous ones. Just avoid that.

5

BTW I’d love to know the motivation for this question since this issue cannot happen if you use mediasoup client side libraries.

1 Like
6

We are using mediasoup-worker alone so I’m trying to read the source code of the C++ part. And this part is pretty annoying for me, i try to find the reason.
At first i think the webrtc will generate the same headerIds for for same key( Chrome always generates 4 for mid), but firefox not, then i think its keep the same in same PC, but still no doc for that.
It bothers me for a while, so i post the question here.
So accroding for your answer, it’s safe to asume in same PC they will use the same headerExtensionID for same key if its generate from a valid client, right?

1 Like
7

The requirement is per Transport so per SDP so per PeerConnection, yes.

1 Like