1531075 - createOffer with no transceivers fails

Status: RESOLVED FIXED

(Core :: WebRTC: Signaling, defect, P3)

Description

[Byron Campen [:bwc]]

This is a degenerate case, but there's nothing in the spec about treating this as an error.

Comment 1

[Jan-Ivar Bruaroey [:jib] (needinfo? me)]

I seem to recall a comment on a spec issue where the Firefox behavior was considered, but I failed to find it. rtcweb-wg/jsep/issues/832 would suggest createOffer wo/transceivers should be supported.

Comment 2

[Byron Campen [:bwc]]

Any gut feeling on priority/rank here? As far as I can see, the primary reason to do this is to improve our test coverage and make wpt sync easier.

Comment 3

[Jan-Ivar Bruaroey [:jib] (needinfo? me)]

No-one has complained. I'd say low priority.

Comment 4

[Jan-Ivar Bruaroey [:jib] (needinfo? me)]

Found it! https://github.com/w3c/webrtc-pc/issues/1745

Comment 5

[Byron Campen [:bwc]]

Bumping priority mostly so we can re-enable some wpt, and ease wpt sync a little.

Comment 6

[Byron Campen [:bwc]]

Attachment: Bug 1531075: Allow offers to be created/set without any transceivers, re-enable some tests, and fix a mochitest.

Comment 7

[Byron Campen [:bwc]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=0dbce8deb3deec75847b0d5baf8aeaa693a45a21

Comment 8

[Byron Campen [:bwc]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=b559b2bfa178ac4f95f33896ae4c0236853274a9

Comment 9

[Byron Campen [:bwc]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c0a578d2155661f3a0da161641df88d43b4ebc8a

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/add69e4bd9af
Allow offers to be created/set without any transceivers, re-enable some tests, and fix a mochitest. r=jib

Comment 11

[Cristian Brindusan [:cbrindusan]]

Backed out changeset add69e4bd9af (Bug 1531075) for wpt failures at RTCPeerConnection-setRemoteDescription-offer.html.

Backout: https://hg.mozilla.org/integration/autoland/rev/5899671d28c9c424f549666af6749ef93da8c92a

Push that started the failures: https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=pending%2Crunning%2Csuccess%2Ctestfailed%2Cbusted%2Cexception&revision=add69e4bd9af8b952a29e66b6cf0df33be47f214&selectedJob=233636814

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=233636814&repo=autoland&lineNumber=20057

Comment 12

[Byron Campen [:bwc]]

Stepping on my own toes here with bug 1531146. Have updated phabricator.

Comment 13

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/7314b493edd3
Allow offers to be created/set without any transceivers, re-enable some tests, and fix a mochitest. r=jib

Comment 14

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/7314b493edd3

Comment 15

[Nico Grunbaum [:ng, @chew:mozilla.org]]

The SDP spec requires connection information be present [0], so this degenerate case isn't expressible with SDP, unless we include a top level connection line. Should the Rust SDP parser be changed to support this case? This seems like a bug in the JSEP spec not to also require a top level connection line in this case, and something that will occur as a developer error (which would be helpful to report), more often than it would be used intentionally.

[0] https://tools.ietf.org/html/rfc4566#section-5.7

Comment 16

[Byron Campen [:bwc]]

I agree that this case is goofy, and should never have been allowed by the spec (either webrtc-pc or jsep). It might be worth surfacing a warning I suppose, but for now let's just allow it.