HTML Media Task Force Teleconference -- 18 Oct 2016

trackbot, start telcon

<trackbot> Meeting: HTML Media Task Force Teleconference

<trackbot> Date: 18 October 2016

scibeNick: plh

<scribe> scribeNick: plh

1. MSE Proposed Recommendation - AC Review closes on Nov 1

Paul: AC review for MSE proposed recommendation closes on Nov 1

EME timeline

<paulc> Tue Oct 25: Completion of EME test suite and finalization of test results

Paul: By next week, we're supposed to have the test suite and have the test results.
... want to make sure we reach that goal
... so that we have to decide the way to go

<paulc> Tue Nov 1: Completion of review of test results to determine features that lack two implementations

<paulc> Tue Nov 8: CfC to publish a revised Candidate Recommendation with features that lack two implementations marked "at risk" (Nov 8-15)

Status of test suite

<paulc> https://lists.w3.org/Archives/Public/public-html-media/2016Oct/0026.html

Paul: Jerry, you talked to a lot of folks. where are we?

Jerry: we had 68 subtests that were failing
... 42 were errors handling errors, ie not returned the proper error type
... Chrome made some improvements on that

<paulc> See Chris's response: https://lists.w3.org/Archives/Public/public-html-media/2016Oct/0034.html

Plh: classify them as bugs in implementation and move on

Jerry: should treat them as non-blocking

Plh: agreed

Paul: Mozilla is fixing those btw

David: for Chrome, I'm checking on things on our end. We fixed a bunch of failure in Chrome.
... don't retest yet
... Firefox nightly should be available for testing
... the crashes should be fixed in Chrome

<paulc> Paul: I have asked Chris to confirm when we can re-test Mozilla.

<paulc> Temporary Session: 6 failures

Jerry: some fixes are in progress
... trying to find a clean list

David: we fixed on temporary thing. key status is fixed.
... changing video encryption is a bit undefined in the spec but no need to fix that v1

Paul: so leaves us with 4 of the temporary

<paulc> 2 of the 6 TS bugs are related Readystate

David: re Readystate, implementations aren't straighforward. we're getting things fixed to make sure we have the correct tests
... if we get the right people on it, like Philipp J.
... 40 is editorial
... not sure how to answer #338

Paul: do we agree there are editorial and non-blocking?

David: editorial is in normative text. for nonblocking, they are state related. we should fix them.

Paul: sounds like 2 of temp session bugs are related to readyState

<paulc> New bugs 340 - Editorial

Paul: #340 is editorial

<paulc> 338 on Timing and it is not clear how to fix this

<paulc> 339 and 336 are under discussion and directly related to implementations of ReadyState

Paul: what's the next critical to fix readyState?

David: getting spec expert point of view

Jerry: some of the temp tests are edge cases
... I'll send an email to follow up on those tests
... core temp scenarios are working. we have edge case failures and persistent isn't working

Paul: Persistent license and usage record

<paulc> Persistent-License Sessions: 11 failures

Jerry: David indicated that a number of those passing but it would give us one passing implementation

<paulc> See https://lists.w3.org/Archives/Public/public-html-media/2016Oct/0034.html for clear statement that Mozilla is:

<paulc> At this time we are not planning on implementing "persistent-usage-record" or "persistent-license" sessions. We won't be trying to make the "Persistent-License Sessions" and "Persistent-Usage-Record" tests pass in the EME v1 time frame.

Jerry: we're passing on 5 DRM persistent test cases in Chrome now

Mark: we can build on top of temporary sessions to support the persistent use cases if needed

Jerry: maybe we can drop those tests?

Paul: I'd rather document why the tests are edge cases

Jerry: I think we have a rational to say that the clear key persistent license aren't important.

Paul: which persistent session will survive, given that they're not implementing
... and Microsoft implementation is outdated

Jerry: I agreed with Ade we would enter bugs related to those tests

Paul: for persistent license today, what do we have?
... we have Chrome
... nothing from Mozilla and outdated implementation from Microsoft

Mark: persistent license isn't implemented in Edge

Jerry: it is

Mark: but I don't think you can access it

Paul: so for persistent license session, we have one implementation, an outdated one, and Mozilla won't do it
... some of the 11 tests are working now
... we could show Edge's plan to bring up their implementation up-to-date

Jerry: the difference between persistent and temporary aren't core and fundamental. these are minor variations in the spec
... so we have an argument for keeping persistent license in the spec

David: re persistent license coverage, I don't expect further changes in Chrome. tests are behaving strangely as well.

Paul: so that would leave us with 5 outstanding failures

Jerry: for temp license, if Chrome is debugged, we'll be all green for Chrome

David: for persistent session clearKey, we could make it optional. temp clearkey would still be there

Paul: persistent usage records?

<paulc> Persistent-Usage-Record: 6 failures

Jerry: we have no implementation that will pass in our timeframe, unless Mark knows otherwise

Mark: other than the polyfill one, you're right.

Jerry: 2 out of 6 pass with polyfill indeed

Paul so, sounds like we'll have to decide real soon how to handle this feature at risk

Paul: reality is that we don't implementation experience with that features

<ddorwin> https://www.w3.org/2015/Process-20150901/#implementation-experience

Mark: depends on what you want to show. we can demonstrate implementability, but not interoperability

Paul: if we only have polyfill support, is that ok?

Jerry: MS has an implementation but not conformant

David: for Google and Mozilla, we're not sure about the feature altogether

Plh: that would make the Director uncomfortable moving forward with it.

Mark: we can keep at risk for now

Jerry: at this point, we probably don't need to republish the CR
... there might be a difference between persistent session and persistent usage session

Mark: we should list what's our criteria for keeping or removing features

<paulc> Other: 3 failures

Jerry: one is clearKey with multiple keys. one is clearkey with update input. one is DRM with multiple keys.
... 2 of them looks like key rotation

David: Chrome has a limitation. don't think it's an issue

Jerry: Chrome passes all 3 of those tests

Paul: what about Edge and Firefox?

Jerry: Chris didn't give specific
... they're working on it

<ddorwin> https://www.w3c-test.org/encrypted-media/drm-mp4-playback-temporary-multikey-sequential.html: assert_false: Should not continue playing whilst waiting for a key expected false got true

Paul: maybe we could take the clearkey with update input out. but looks like we're ok in the misc category
... so we're waiting to know when to retest
... also waiting for Edge bugs
... need to decide what we need to do with the persistent cases
... whether we're dealing with intent or problem to implement

Mark: we still have some problems with the expiration of key state attribute

David: the expiration is more than expected. other test is strange, might be a chrome bug.

Mark: worked ok on chrome but don't think it works well on others

Paul: what are those failures in the summary?

Jerry: not sure. will have to look back.

<ddorwin> http://w3c.github.io/test-results/encrypted-media/all.html

<ddorwin> shows two expired/expiration tests

Mark: missing a test for license renewal
... for expiration time, we have some clock skewed issues

David: I'll look at the license renewal test to see how to trigger it

Jerry: drm-expiration passes in Ch and FF

test suite issues and pull requests

Paul: I think we're good here.

<ddorwin> https://github.com/w3c/web-platform-tests/issues/3982

<ddorwin> Test does not match spec.

<paulc> https://github.com/w3c/web-platform-tests/issues?utf8=%E2%9C%93&q=is%3Aissue%20is%3Aopen%20label%3Aencrypted-media%20-milestone%3AVNext

David: we need to fix the test

Mark: I'll look at it

<paulc> Test suite pull requests: https://github.com/w3c/web-platform-tests/pulls?utf8=%E2%9C%93&q=is%3Apr%20is%3Aopen%20label%3Aencrypted-media

<paulc> https://github.com/w3c/web-platform-tests/pull/3985 is open

David: Chris needs to reply on 3985

Mark: shouldn't be a problem to fix up that test

https://github.com/w3c/web-platform-tests/pull/3936

Mark: that's a new test
... don't remember the implementation status but the test needs review

David: won't be able to review this week

Paul: I'll see if Chris can look at it

Next steps

Paul: most important item is to rerun the tests. we'll need a meeting to see where we are
... also need to decide if we need a CR or a PR
... hoping to make it next week

- DRAFT -

HTML Media Task Force Teleconference

18 Oct 2016

Attendees

Contents

1. MSE Proposed Recommendation - AC Review closes on Nov 1

EME timeline

Status of test suite

test suite issues and pull requests

Next steps

Summary of Action Items

Summary of Resolutions

Scribe.perl diagnostic output