SdpInfo.js wrong behavior in simulcast

[kekkokk]

Hi,
I'm facing some issue with firefox Simulcast.
I think that there's something handled in the wrong way in the new SDPInfo.js
#1075

Before this push, everything worked fine. We made few adjustments in c++ to make firefox simulcast works.
(Client side) I'm calling the .enableSimulcast() function in the SdpHelpers BEFORE creating the offer. Firefox then generates an offer with:

a=simulcast: send rid=spam;egg
a=ssrc:1099767458 cname:{299dbffc-d388-8644-b3d2-8974094ff9df}
a=ssrc:1589187499 cname:{299dbffc-d388-8644-b3d2-8974094ff9df}

then setLocalDescForOffer is called and then the sdps flows into SemanticSdp.SDPInfo.processString:

that.localSdp = SemanticSdp.SDPInfo.processString(that.localDesc.sdp);

The problem is that the

a=simulcast: send rid=spam;egg

directive disappear.

THIS is the difference between the generated sdps from the browser (untouched) and the processed sdps from SdpInfo.
As you can see the a=simulcast get lost and there's also duplicate

a=ssrc:1589187499 cname:{299dbffc-d388-8644-b3d2-8974094ff9df}
a=ssrc:1589187499 cname:{299dbffc-d388-8644-b3d2-8974094ff9df}

[jcague]

sorry, we don't support Firefox simulcast currently in the repository, so I can't help. But you have two options: you can find the bug and submit a patch to the SdpInfo or, even better ;), submit the code that makes Firefox simulcast work!

[kekkokk]

I can push the code tomorrow with no problem, I submit everything I do (see the update to latest libnice).
But It will not work cause of the issue. If you can point me in the direction to comprehend why the a=simulcast line gets dropped, I can solve then PR.

[jcague]

sure! probably I can fix it in the same Firefox Simulcast PR with the rest of the code if you prefer.

[kekkokk]

Still, I see the duplicate

a=ssrc:1589187499 cname:{299dbffc-d388-8644-b3d2-8974094ff9df}
a=ssrc:1589187499 cname:{299dbffc-d388-8644-b3d2-8974094ff9df}

cname as a bug.

I mad simulcast works in the last commit of my PR but the layers are inverted. layer 0 is 640x480 and layer 1 is 320x240. I think is due to this bug. I'm still investigating