Notes, April 21, 2016

On 04/21/2016 10:43 AM, Harald Alvestrand wrote:
> Overarching items
> ============
> CR status for mediacapture-main: The chairs have decided to go ahead
> with a request for CR status, and expect to send the request on Friday.
> We'd like to close #333 and #338 by merging #340 and #342 and produce
> another editors' draft as the basis for the CR request, but given the
> way the process works, Dom says we can do that in parallel with
> actually sending the request.
>
>
> Mediacapture-main
> =============
NOte: Travis seems to error out on warnings. Should it?

> Pull requests
> ----------------
> #340 Restrict emission of devicechange event to apps with list-d ()
> Good?

Merged.
>
> #341 Make ended event a simple event ()
> Merge?

Conflict in the changelog. Merge-ready.
Suggestion: Tell people to suggest a changelog entry in the PR rather
than adding it themselves.
>
> #342 Remove obsolete table of error names ()
> A bit sad, would be nice to have a (non-normative) list of the error
> names we use, rather than the names we define. But should be OK.

Merge-ready. We can make a table later.
>
> #343 Various fixes to permission language ()
> Merge?

Merge-ready.
>
> #344 Add [CEReactions] annotation to allowUserMedia ()
> Seems right if you buy the premise that these annotations are a
> reasonable way to go. But not HTML 5.1?

Not going to merge it at this time.

>
> #345 Issue 316:  add links to constrainable properties ()
> LGTM, merge?

Merged.
>
>
> Issues
> ---------
> #316 MediaTrackSettings/ConstraintSet/Capabilities... should lin
> (burnburn)
> Fixed by #345
>
Closed.

> #333 Devicechange event firing potentially being a fingerprint (stefhak)
> Fixed by #340
Self-closed.
>
> #334 Is more needed regarding revocation? (npdoty)
> Still open, I think. npdoty commented 2 days ago. Assign?

Callback capability in permissions API? Stefan to take further
>
> #338 Make ended event a simple event (dontcallmedom)
> Fixed by #341

To be closed when PR merges.
>
> #339 originIdentifier is unused (adam-be)
> "Permissions storage identifier" is still a permissions API term. We
> shouldn't redefine it.

Adam will provide a fix.

Stefan took on the task to suggest removing the SourceType in a bug, and
(perhaps) make a PR for it.

We need to get a new draft out. Dan took on the task to create a new
one, ping Cullen and push it.

>
> Webrtc-PC
> =======
> Pull request
> -----------------
> #466 Use an enum to describe directionality of RTP Stream (alvestrand)
> #467 Use enum for voiceActivityDetection (alvestrand)
> #471 Use enum for RTCDataChannel’s ordered attribute (alvestrand)
>
> Proposal is to say "merge, close, close". No comments on list yet.
> Wait a week.
>
> #580 webrtc.html: misprint and basic fixes ()
> Looks friendly. Merge?

HM: Travis has not checked IPR on this guy. Needs checking.
Dom says it's OK. Merged.


#587 info on MediaStreamTracks in PC
Assigned to Bernard.

#589 copy/paste error
Merged

#590 nits from Dom
He's usually right. Merged.

>
> Issues
> ---------
> Suggestion: First start at #551 and assign things, then work through
> the assigned items.
>
> #142 "Hold" unspecified (aboba)
> #253 Assurance that requests to IdP proxy originate from the use
> (martinthomson)
> #257 ICE Candidate should have accessors for protocol-relevant i
> (alvestrand)
> #295 Guidance for extending objects vs extending Stats needed (alvestrand)
> #296 Debugging ICE problems needs more info (aboba)
> #305 Describe what happens when media changes (fluffy)
> #337 Interfacing between WebRTC spec and JSEP (burnburn)
> #369 addTrack's streams parameter is unused. (adam-be)
> #375 true as default values for dictionary is bad practice (alvestrand)
> #440 Use interfaces not dictionaries (adam-be)
> #457 Non-normative ICE state transition diagram (aboba)
> #517 Clarify ICE consent freshness feedback (aboba)
> #518 PING review of webrtc-pc spec (stefhak)
> #526 NetworkError event is not defined and might not be needed (adam-be)
> #529 MediaStreamTrack clarifications needed(?) (stefhak)
> #542 Server cannot be reached - Issues with IPv6 (aboba)
> #548 RTX/RED/FEC handling (aboba)
> #550 'the process to apply candidate' should link to JSEP (adam-be)
> #551 Errors when identifying a m-line in addIceCandidate() ()
Link to 3.5.2.1 in JSEP? (MID has precedence - bogus MID should be an error)
File a bug on JSEP to be clear about that.
> #554 We never fire the 'connectionstatechange' event ()

Need a section similar to "life of an ICE agent" to handle "monitor the
input variables".
we should have a general way to describe this. Adam to think about.

> #555 Sort out requirements around IdpLoginError (martinthomson)
> #561 Normatively cite webrtc-stats for sections 8.x ()
No.
> #562 What to do with an RTCIdentityProvider that returns rubbish
> (martinthomson)
> #563 Undefined 'transceiver' object used in 'dispatch a receiver ()
Pthatcher.

> #565 pc.close() never fires any events. On purpose? ()
Not firing events is consistent with the track.stop() behavior - the
application caused it, so doesn't need to know.
Stefan will bring it to the list.

> #566 Separate sender and receiver sets are unnecessary when we h ()
Agreed, and useful to clear up. Dan takes it, Adam will review.

> #568 Should we specify how addStream()/"addstream event" should  ()
Polyfilling addstream() is easy; polyfilling the addstream event is
really hard.
We need backwards compatibility. But with browser implementations or
with the previous version of the spec?
Bernard will take care of doing cat-herding.

> #570 RTCRtpReceiver.track needs to be nullable ()

Adam will take the underlying issues forward; closing this one.
(and that was the top of the hour.)
> #571 Mechanisms for populating the contents of RTCRtpSender/Rece ()
> #572 Unclear how the `getStats` track selector works in the cont ()
> #578 Need to specify precisely when MID generation happens (adam-be)
> #579 Congruenting about "The negotiation-needed flag is cleared  ()
> #581 No error defined for addTransceiver() with a bogus track ki ()
> #582 Typo in addTransceiver() ()
Fixed by #589
> #583 Is it OK to call addTransceiver() with a track already adde ()
> #584 currentLocalDescription: incongruous explanation ()
Fixed by #580
> #585 Unclear if RTCRtpTransceiver.stop() acts right away or requ ()
> #586 The InvalidParameter error seems to not exist ()
Closed.

>
>
>