STRs:

1. Open https://jsfiddle.net/jib1/qcwe1t86/show and share camera

Expected result (<77):

• Three smooth-running live camera videos in small, medium, and large size

Actual result:

• Whole page flickers epilepticly, as if video dimensions are rapidly flipping between different sizes
• The first (small) video somewhat works, but updates choppily (~4fps)
• The second video (medium) appears frozen in low res; same size as small. On some runs observed to be at half brightness
• The third (large) video is large but similarly frozen

Nils pointed out in #media that this fiddle is a hack that tries to receive simulcast: It splits one m-section with one MID and three SSRCs into three m-sections. These three m-sections will still have different SSRCs, but all the same MID.

I guess what remains to be determined is if this fiddle foreshadows any problems with any SFUs out there. So far I haven't seen any.

I am already seeing signs of something broken in simulcast over in bug 1636185.

Some related good news, fippo's new rid<->mid simulcast trick used in WPT works fine in Firefox, modulo bug 1401592. 🎉

So I guess I'm less worried now since we have a path toward testing simulcast. I don't know about comment 2 though.

The support for receiving simulcast (which is not expected for browsers) was added to support testing simulcast. It should still work when MID is not employed. This might be fixable to work with RID based filtering.

Jan-Ivar, can you determine a severity for this bug? The last comment makes me suspect this is S3 or even less?

Byron, WDYT?

Certainly no more than S3 I think. This recv simulcast stuff is a hack, really.

Wontfix for 77 given that it is S3 and we already shipped.