See also: IRC log
<scribe> scribe: nigel
Nigel: Next week I'm able to make the call, but the week after, the 12th, I can't.
Cyril: I also can't make the 12th.
Pierre: Regrets for me for the 19th.
Nigel: For today, we have TTWG
Charter, if there's more to discuss on that - will wait
for
... staff to join before confirming.
... We also have TPAC, and a possible F2F in Munich in
May,
... TTML1 3rd Edition CR needs to be discussed.
... Not sure if there's anything to discuss on TTML2.
... For IMSC there is one agenda point, a pull request, which
we may be able to resolve with
... a brief conversation.
... If we have time we can go through the IMSC vNext Reqs pull
requests, which have been
... open for a while.
... I don't think there's anything to discuss on CSS.
... Do we have something for WebVTT?
David: Yes, we should approve the transition to CR.
Nigel: Any preferences about what order we do these in?
Pierre: I'd really like to close
the IMSC ticket, because it is blocking CR.
... If there's anything missing on that push-back on TTML1 we
should address that ASAP.
Nigel: We should be able to get through everything today - we're scheduled for 2 hours.
Nigel: Thierry wants us to
discuss if we wish to meet at TPAC in Lyon, which is at the
end
... of October this year.
... I expect us to be getting towards the end of our Rec
transitions for all our specs at that
... time, but I expect there will be a lot to discuss, so I
propose that we ask for what we
... usually ask for, i.e. 2 days.
Pierre: Sounds good.
Nigel: Okay, I'll complete the
WBS for TTWG asking for that.
... There will be a request for any groups we don't want to
clash with, or want to have joint
... meetings with. I would at least propose that we ask for a
joint meeting with CSS WG like
... we did last time, on the basis that we expect to have made
some progress.
... Any other thoughts?
group: [silence]
Nigel: The next proposal is that
we have a f2f on May 22 and 23rd at IRT in Munich.
Andreas
... has kindly offered to make space available there before the
IRT subtitle technology symposium,
... and I think the timing will be good to iterate over TTML2
and IMSC 1.1 tests for the test suite.
Cyril: I might be able to make that.
Pierre: It's unlikely I'd be able to make it.
Nigel: Okay, I'll send an email
round about this; for now it is a proposal. I think if we
... intend to talk about test suites we need the right people
in attendance.
Nigel: We submitted the
transition request to CR for TTML1 Third Edition and
Ralph
... responded expressing surprise that no new tests have been
created to verify that implementations
... conform to the clarifications and error corrections. He
asked if the test suite has been
... updated at all. He's basically asking us to update the test
suite to demonstrate it.
Pierre: I think it only needs to be updated selectively to cover the areas changed.
Nigel: I think that's right.
Pierre: I think that's possible to do.
Nigel: I think if we're claiming
that implementations meet the updates already then we
need
... to provide evidence.
Pierre: My recommendation is to
change the SOTD to include exit criteria that require
... passing tests, and then create those test suites.
Alternatively I could provide GitHub
... pointers to issues on imsc.js, but that might be a lot of
work with no guarantee of success.
Nigel: That would work.
... If we're going to do that we need to make the updates and
re-file the transition request.
Pierre: I'll make the SOTD updates.
Nigel: When that's done, please let me and Thierry know so Thierry can update the transition request.
Nigel: we have one pull request on the agenda.
github: https://github.com/w3c/imsc/pull/350
Nigel: I think the main issue
here is the syntax permitted for tts:extent.
... I don't understand why the `auto` value is permitted for
`tts:extent` on `tt:tt`...
Pierre: It was permitted before.
Nigel: But has its meaning
changed? It used to be defined as the Root Container
Region,
... now it is defined as "contains" whose meaning is defined by
appendix H.
Pierre: When I went through this
before I decided that "contains" ends up meaning the same
... as "auto" used to mean in TTML1 - it yielded the right
outcome.
... Note that in IMSC pixelAspectRatio is prohibited, so
"contains" resolves as the display
... aspect ratio of the root container region.
Cyril: And the algorithm in H.1 and H.2 match?
Pierre: Yes, if there's no DAR
and PAR then it's H.1.1 which corresponds to TTML1 100%
100%,
... i.e. auto semantics.
Nigel: So Pierre you're arguing
that contain and auto resolve to the same so it is better
to
... have a single syntax option rather than allow the
semantically more precise "contain"?
Pierre: Yes.
Cyril: There may be cases where auto and contain resolve differently?
Pierre: Outside the tt element, maybe, but on the tt element they are identical.
Cyril: Yes, the spec says if the
value is auto, and its the tt element, then it maps to
contain.
... So they are equivalent.
... I think the constraint on IMSC is correct.
Nigel: Okay, I'm persuaded.
... The next comment here is about extends and restricts.
Pierre commented that
... there's no requirement for them, and presumably there's no
benefit identified for them here?
Pierre: No, I have not.
Nigel: I can see that from an
implementation perspective it may add a fair amount of
code
... and tests to allow extends and restricts.
Pierre: Also this new TTML2
syntax was not broken out as a specific feature
designator.
... I think I raised an issue on TTML2 for that.
... Maybe later it will be made a feature and we can then
refactor this text.
... It would be a lot clearer.
Nigel: Ok, there are conflicts here but I'm happy to approve the changes.
SUMMARY: @nigelmegitt to approve the pull request
Pierre: The conflicts are super-boring, I'll fix those.
github-bot, end topic
github: https://github.com/w3c/imsc/pull/354
Nigel: You're waiting on someone from i18n on this, Pierre?
Pierre: Yes, during the call
people used the term "safe" but the comments were against
that.
... Instead, I've proposed to use "common" and am waiting for a
response on that.
Nigel: That change was made on
the pull request, but there's no comment about it, but
... there is on the issue?
Pierre: Oh yes, I could prompt Addison to review with a comment on the pull request. I'll just do that.
Nigel: I'm not sure what we can do more on this at the moment.
Pierre: It is only a matter of finding the right terminology now, I think we're good on the rest.
Nigel: Looks that way.
SUMMARY: WG awaiting review feedback from i18n.
Pierre: We should communicate to
them that we plan to request transition to CR in 7 days
... so they need to come back soon if they have a strong
concern.
Nigel: That sounds like an action for me to send a message immediately after this meeting.
github-bot, end topic
Thierry: Not much to discuss - I
have submitted the draft charter to W3M, and I think they
... were supposed to review it yesterday but I have nothing to
report yet.
<scribe> Chair: David
David: At TPAC we looked at the
transition request, and the actions requested have been
... done. We were waiting on closing out some issues on the
spec which Nigel was unable
... to get to for the first few weeks of this year. My feeling
is that the remaining issues can
<dsinger> https://lists.w3.org/Archives/Public/public-tt/2018Mar/0133.html
David: be closed off in CR. I
would like to ask for the formal agreement of the group to
do
... the final transition. I sent the final draft of the request
a couple of days ago.
... I updated the wide review page a couple of days ago to
reflect the current status. I hope
... we're at the formal stages now - we had the [scribe loses
track]
... There will likely be some changes to be made during CR,
which are not major changes
... for implementors but may for example require a change to
the computed CSS property value for something.
... I think we need an updated version of the spec with SOTD
etc for CR, which Silvia and/or
... Thierry can do.
<dsinger> The remaining edits are clarification, warning, and so on and don’t represent technical changes to the specification, so I think it’s safe to do them along with any other disambiguations needed by implementers during CR.
Nigel: What I normally do at this point is ask Thierry what is needed now for the transition request?
Thierry: For the SOTD there are a
few things that need to be put in, first the exit
criteria.
... From David's statement, that is like what we have been
using within this group, 2 implementations
... for each feature, so that sound good.
... The second thing I have not seen are the features at risk,
because there are some features
... that are not implemented like regions and some others.
David: We discussed this, it's at
the end of point 1, and as discussed, we don't want to
drop
... them so we aren't marking them as at risk. They are not
features to drop if they are not
... implemented. The group wanted to wait until they're
implemented, and have no features at risk.
Thierry: Okay. Another thing is
for the test suite and the implementation report. Of
course
... we have to fill the implementation report in later. We
should have a link to a test suite
... or something if it is incomplete.
David: Yes, that's also in the
transition request, there's a fairly thorough test suite
in
... web platform tests, which generates automated reports for
browsers, and we're going to
... have work out how to do that for non-browser
implementations during CR. That's for
... me and the group to do during CR.
Nigel: We don't normally ask for that at this stage?
Thierry: We don't need the whole thing, just a link to whatever is there.
David: We believe it is pretty
thorough at this point, I'm sure there are bugs that people
will
... find during implementation work.
Thierry: The last bit is the wide
review, I guess it is done, we have a URI, and it will be
up
... to the Director to review it.
... The last thing is WG approval for transitioning.
... I need a link to point to.
PROPOSAL: Agree to the CR transition to WebVTT
Pierre: For clarity, that's based on the gh-pages branch on GitHub?
David: Yes, that's correct, Silvia will need to update that.
Pierre: And all the licence and
IPR issues have been addressed and there's not going to
be
... any drama there?
David: I don't think so, no
exclusions have been filed so far and we asked for FSA from
all
... the CG contributors.
Pierre: And some of the
contributions have been from the CG since then, or all from
members of the WG?
... (after that commitment was received)
David: Even if they came from the
CG they did sign a CLA. The only issue would be if they
... contributed someone else's IPR and that's a door I don't
know how to close. There's nothing
... that's giving me any anxiety.
Pierre: So the next step is to
create a CR branch for the group to review the final
version.
... When will that be available?
David: You'd like a formal branch in GitHub?
Pierre: Normally that's what
happens, so there's a formal document to review.
... I'm encouraging you to do this as soon as possible so there
aren't any surprises?
David: I'll see what we can do - is there anyone on the team who can help?
Thierry: I can help on the SOTD of course.
Pierre: I don't have any a priori objections, thank you for doing all this additional work.
Nigel: I'd second Pierre's request, let's have the actual document that we are going to approve on the table.
Thierry: It has to be a CR version for approval by the Director.
Pierre: And to be clear, on the open issues you do not expect any formal objections?
David: I'm not seeing any clouds on the horizon for the remaining issues.
Nigel: Me neither.
... David, you've been clear that you have limited resource for
chairing and editing. Will there
... be the effort available to make any changes and to work on
it post-CR so that it can get
... to Rec? My concern is that it could linger in CR
forever.
<dsinger> PROPOSAL: The editor and the team to prepare the CR. The WG approves the CR transition, with said prepared CR to be presented to the Director for approval, using the transition request in https://lists.w3.org/Archives/Public/public-tt/2018Mar/0133.html
David: I share your antipathy to
specs staying in CR indefinitely, I would suggest that if
... we cannot get to Rec within the next Charter period then at
that point we publish the
... spec as a Note and stop working on it in the WG.
Nigel: Thank you, by time-boxing the CR that addresses my concern.
Pierre: I think you can
confidently create a CR ready spec and state a resolution to
proceed
... with CR with this draft, intended to be the CR.
... Then there's no surprise. Otherwise we could take a
resolution today, and in 2 weeks there's
... something objectionable to someone and then we restart the
clock in 2 weeks.
David: I was expecting the group
to approve transition today. Thierry, can you prepare the
... CR version of the document today?
<dsinger> Can Thierry do the pull request for the SOTD section in the next 24 hours?
Thierry: I can work on it, and
probably not in the next 2 hours, but tomorrow morning.
... We also need in the SOTD the date for earliest advancement
beyond CR. Probably we can
... put 3 months or whatever.
Nigel: Good point.
Thierry: I propose at least 2 months.
David: It's not going to happen
for at least 6 months. We need implementations of the
changes
... and of regions. Give it 6 months.
Thierry: You don't need to do that, you can do it in 3 if those criteria are met.
David: Okay, put 3 months then, that's fine.
Philippe: Hello, I apologise for
arriving late.
... I don't mean to derail this meeting, but I have one
question. We have not started review
... of the TTWG charter already, but I have one question: How
is WebVTT used on the web
... today? Not an implementation question, a usage question. Is
it actually used on the web,
... or only as an input format so video services can do their
own thing with captions.
... TTML is used as an input format, and then there's client
side JS that takes that and displays
... the subtitles and captions at the right time. YouTube,
Netflix and others don't use TTML or WebVTT,
... they use their own code to present the captions.
... So you don't need native implementation of captions.
... The second question is how are we going to be able to get
out of CR for WebVTT?
... If my assumption is correct, there is no incentive for
browser implementers to update
... their implementations. That's part of the questions that we
are asking ourselves generally
... about the future of captions on the web.
David: I agree a lot of captions
are done with polyfills and HTML/CSS created on the fly.
... One of the issues with WebVTT is that it is not used for
presentation.
... I do know that in [apple products] we take the WebVTT
natively. [sorry, scribe subject to a lot of local background
noise]
Philippe: Ok, I didn't find any
statistics, for example in Chrome, of how often the
native
... implementations are used today.
David: Okay, I'll try to find out.
Philippe: The fear is that it is not used so we will never get to the top of the priority list for browsers.
David: I share your concern, if
you can do an adequate job with polyfills then who needs
to
... do a native implementation.
Philippe: Yes, that doesn't
undermine it as an input format.
... For example I do not know if the WebVTT implementation
natively would allow positioning
... of captions on the fly like YouTube allows.
... We may be chasing something that the market out there is
not interested in having in terms
... of native implementation.
David: Right. Yes.
Nigel: In terms of resolutions, I
think the next steps are to produce the CR version of the
... document and then for David as Chair to send a message to
the group specifying the
... resolution that he believes has been made, and highlighting
the review period under the
... group charter decision policy (which is 10 working
days).
Philippe: At the moment we could
try to run the transition request in parallel as long as
it
... is clear to the Director that there is an opportunity for
the decision to be reversed.
... On the IPR question, if there's a question about CG/WG
working, here it is the same
... as if a contribution comes in on the WG public mailing
list, where we have to figure out
... how substantive the contribution is and address that. It
doesn't change the risk
... associated with IPR in the spec that is not directly from a
contributor. If you have concerns
... about that then we can engage with our legal team here and
make an assessment.
... If you tell us that everyone who contributed is in the CG
and the WG then we don't have
... an issue.
Philippe: A comment - the
agreement you described before is that if you do not have
Rec
... of WebVTT by the end of the next Charter then you would
drop it. If the theory is correct
... that WebVTT is used as an input format rather than a native
implementation then it would
... be no surprise if that happens.
David: One question is if you
would accept two implementations from Apple as being
... independent, because this is in fact the case.
Philippe: Interesting question, I don't know the answer but I can ask and get back to you.
Nigel: I have asked this recently, as an issue on the Process.
Philippe: Yes, you would need to
have some evidence that the two teams creating the
implementations
... did so by interpreting the specification directly without
any other communication.
Nigel: Any other questions or comments on the Charter?
Philippe: Not as far as I know. Thank you by the way for providing the draft Charter.
Philippe: I'm in communication
with Travis to try to make the pull request smoother.
... I do not know why the repositories have been deactivated.
We are reaching peaks of 60 concurrent
... jobs on travis at the moment and it keeps increasing. They
are doing some of our jobs
... in batches and they can get delayed before they are even
started. So there's both a delay
... and then a long queue before they run. So the plan is to
conduct an experiment on travis
... to give them a bit of money to provide small guarantees and
see how it affects our jobs
... being run. In parallel I've been talking to the web
platform tests people because they are
... consuming more than half our jobs, just for testing
purposes, and we cannot separate
... them because they are on the same organisation on GitHub.
We're potentially considering
... moving them to a "separate" organisation on GitHub because
that project is going to
... grow more and more. Then we can separate testing from
production of recommendations
... more easily. Every time a pull request on WPT is done it
triggers 12 concurrent jobs.
Nigel: Thanks for that.
Nigel: I've been in discussion
with various organisations (we're up to 12 right now)
about
... setting up a CG to produce a profile of TTML2 for audio
description, and hope that will
... go ahead in the next few weeks.
... Just noting it here in case people want to join.
Pierre: That's in a CG not the WG?
Nigel: Yes, and hopefully bring
it to a future iteration of the TTWG Charter when there
is
... a document to work on. This is partly about WG mechanics -
getting onto the Charter
... without a document as the basis of a specification seems
harder these days, so this way
... we can have easy participation from the interested parties
and then there's a path towards
... getting to Rec later, albeit one that requires more W3C
membership in the case that the
... contributors are not currently members.
Pierre: And the domain is all applications?
Nigel: Yes, not only broadcast,
also web, and not making any assumptions about where in
... the distribution chain any audio mixing might happen.
Pierre: Thanks.
David: What I did with WebVTT is
I got a FSA signed by the CG participants - if you're
... expecting them to give IPR away for free then they don't
get anything back.
Nigel: Right, and I've highlighted this to the participants right from the beginning.
David: It took me months to get this for WebVTT.
Pierre: Yes, when the legal team looks at it things could take longer.
Nigel: Thanks everyone! [adjourns meeting]