- From: Jan-Ivar Bruaroey <jib@mozilla.com>
- Date: Fri, 11 Oct 2019 23:35:28 -0400
- To: public-webrtc@w3.org
- Message-ID: <66686685-adf0-91bd-2371-55e242ec071a@mozilla.com>
This is the summary of decisions reached at the TPAC meeting of the
WEBRTC WG in Fukuoka, September 2019.
This summary is brief in style, and meant to be read in conjunction with
the agenda, slides and minutes stored here:
https://www.w3.org/2011/04/webrtc/wiki/September_19-20_2019
All decisions are subject to review on the mailing list. For decisions
where no further comments are heard on the list, we will assume that the
decisions listed here stand. To comment, please reply to this message or
start a new thread.
Decisions:
*
WebRTC-PC
o
Send nothing on disabled audio track, regardless of CN/DTX
(since audio has no "frozen picture" problem).
o
Remote tracks have width, height, frameRate & aspectRatio
readonly getSettings(). OverconstrainedError on applyConstraints().
o
RTCPeerConnectionIceErrorEvent hostCandidate must be "" or null
in mDNS case; new port member
o
Keep the ability to postMessage RTCCertificates; add a security
note (because it may be useful for datachannels in workers in
the future).
o
Features at risk:
+
Move OAuthCredential to an extension spec
+
Remove negotiate value from RTCRtcpMuxPolicy
+
Remove VoiceActivityDetection
+
Remove getSupportedAlgorithms
+
Move SendParameters.priority, RTCDataChannel priority and
RTCPriorityType enum to DSCP extension spec
+
Remove dtx, ptime, and codecPayloadType from
RTCRtpEncodingParameters
+
Migrate the remaining identity steps to webrtc-identity
*
Screen capture
o
“A browser display surface is the rendered form of a single
s/document/browsing context/” (to clarify tab capture).
*
Media capture main
o
Broad interest in proposal to hide more device info before
device-info permission is granted (see notes for more).
o
Broad consensus that deviceIds must be double-keyed and not
single-keyed when a page’s storage options are all double-keyed
(see notes for more).
o
Do not fire devicechange event if list of devices is unchanged
(unless doing so is privacy-neutral)
o
Add: “The media MUST NOT be stretched or have fake data created
that did not occur in the input source.”
*
Media capture record
o
Add note that implementations may avoid re-encoding remote
tracks only when mimeType is unset by JS.
o
Proceed with recorder.replaceTrack(a, b) proposal, but make it
promise based (need to discuss multiple track changes).
*
WebRTC stats
o
Remove onstatsended and continue discussion about an event to
notify when significant stats change.
o
Move track stats to obsolete section.
o
Move RTCMediaStreamStats to obsolete section.
o
Add rid and confirm that we don't want to add encoding parameters
o
Add a stat which is the sum of the square of inter frame intervals
o
Audio stats are not per-channel. Use any one channel to
increment counters; use highest totalAudioEnergy; use lowest
echoReturnLoss.
o
Add RTCTransportStats.iceUfrag (to distinguish new candidates
after ICE restart).
o
Look to add “sctp-transport” with some cumulative stat to use
for roundTripTime, but otherwise rely on spinfo_srtt
o
Don't add SCTP bandwidth stats but discuss adding a dedicated
API (not in 1.0)
*
Content hints
o
Advance but at low priority
o
Rewrite draft to include MUST/SHOULD and make track.clone() copy
hint.
*
DSCP API
o
Move Priority to DSCP document, harmonize, keep experimenting.
*
WebRTC ICE
o
No opposition to adopting the spec as a working draft.
*
WebRTC SVC
o
Remove drawings from SVC spec; AV1 is normative for drawings.
SVC spec is normative for strings
Notes and actions:
*
WebRTC PC
o
Developer feedback:
+
Stuck in Plan-B. Please revive SSRC signaling draft to aid
transition to Unified plan.
+
Please see minutes for additional feedback.
o
ACTION:Florent to investigate upstreaming his simulcast loopback
test page as WPT test
*
Screen capture
o
Needs horizontal review, fill out privacy and security
questionnaire. Need explainer.
*
Media capture main
o
Broad interest in proposal to hide more device info before
device-info is granted.
o
Some interest in proposal to go further and lie about the number
of devices for privacy.
o
Broad consensus that deviceIds must be double-keyed and not
single-keyed when a page’s storage options are all double-keyed.
o
Some interest in double-keying deviceIds regardless (though
concerns with this have been raised since TPAC).
o
Interest in a way for webdriver to add/remove/setup web capture.
PR welcome.
*
WebRTC stats
o
Broad interest in removing or gating networkType. Remove “vpn”
value.
o
ACTION: Youenn to upload simple validation tests from webkit to WPT
o
ACTION:Henrik to add PR for SVC stats
*
WebRTC SVC
o
ACTION:aboba to add a paragraph to the table to describe
considerations for adding new values to the table of modes.
*
WebRTC NV Use Cases
o
Split conferencing PR in two. Consensus to merge the untrusted
JavaScript use case. Jan-Ivar will go back to Mozilla to clarify
if we can merge the trusted JavaScript case, where there is
disagreement.
*
Rechartering.Discussed options, no clear answer, but overall feeling
that the WG needs to keep operating to manage post-Recs of WebRTC
1.0 and getUserMedia, finish its other existing specs. For new
features, the question was raised on how they should be incubated.
Will discuss further.
Received on Saturday, 12 October 2019 03:35:34 UTC