Timed Text Working Group Teleconference

<scribe> scribe: nigel

This meeting

Nigel: Looks like we're low on numbers for today, let's see what we can do...
... I don't think we can do much on TPAC, there's a bit on CSS features, there's an IMSC
... issue on origin and extent, and some imsc-tests changes.

Pierre: I'd like to talk about where to host IMSC2 requirements - either on the imsc repo
... wiki or in a different repo. I don't know the lead time to creating a repo...

Nigel: When plh does it, it's very quick, but let's think about if we need a new repo.
... Let's cover it in the IMSC section. For TTML I need to report back on the privacy meeting
... from last week.
... On WebVTT I think David considers that we've had long enough to review the CG's
... proposed disposition of comments, so let's check that's true.
... Any other topics to discuss?

group: No other topics.

CSS styling equivalents of TTML2 and IMSC features

Nigel: There's been one change to the wiki page since last week, because someone from
... CSS WG pointed out that textOutline maps somewhat to stroke-width in CSS.

CSS Requirements - textOutline

Nigel: I changed it from no mapping to partial mapping. This is one where the CSS spec
... is in WD and maybe FPWD but the browsers all support it according to caniuse.

Pierre: I didn't find that when I did imsc.js, but great news.

Nigel: I haven't experimented with seeing if text-shadow can be used alongside stroke-width
... to generate a similar effect.

CSS stroke-width property.

Pierre: [let's try it...]
... It looks like stroke-width is supported by Chrome but not stroke-color.

Nigel: caniuse says they are both supported with a -webkit- prefix.

Pierre: That works, with the prefix on both.

<pal> https://codepen.io/palemieux/pen/ZJpwxJ

Pierre: But it looks like the stroke is internal rather than external.
... It looks like the stroke is centered on the glyph outline rather than going outward,
... so for use in subtitles it is not useful.

Nigel: The comment was at https://lists.w3.org/Archives/Public/www-style/2017Jul/0030.html

Pierre: I'll reply to that thread.

Nigel: Thanks. The next point on this topic is that I'll be dialling in to the CSS WG tomorrow
... 0930 Paris time for their discussion of the styling requirements we have here, and TTML2
... more widely.
... My goal will be to raise the requirements that we have identified and hopefully get
... some agreement to work on them so that we can follow up at TPAC. I'll also invite them
... to attend our meeting at TPAC also.

IMSC

Pierre: I'm interested in understanding the best place to gather IMSC2 requirements. One
... option is the IMSC repo wiki. It's not ideal for change tracking, but maybe we won't need
... that very much. It's super simple. Another alternative is to create a repo called e.g.
... imsc-vnext-req and do it there.

Nigel: I slightly prefer a separate repo and document because that allows us to reference
... it in the transition questions later, instead of saying "no requirements, just based on a
... member submission" we can say "see WG Note ..." which is helpful.

Pierre: Especially if we are going to publish as a WG Note then doing it in a new repo is
... a no-brainer so we should just create a new repo for that.

<scribe> ACTION: tmichel Create a new repo imsc-vnext-reqs [recorded in http://www.w3.org/2017/08/03-tt-minutes.html#action01]

<trackbot> Created ACTION-500 - Create a new repo imsc-vnext-reqs [on Thierry Michel - due 2017-08-10].

github: https://github.com/w3c/imsc/issues/249

Nigel: [summarises most recent posts on the issue]
... Is this just something that followed through from TTML1 without anyone noticing?

Pierre: Exactly, I think it's that simple that nobody noticed it so the ability came through
... without anyone constraining it. The "either" shouldn't be interpreted as a constraint.
... The most we could do is add a Note saying not to do it, but actually supporting mixed
... units is not a problem in implementations regardless of how strange it would be to use them.

Nigel: I think the answer is that mixing units is allowed.
... I don't even think we need a note on it.

Pierre: We could remove the "either" to remove the hint to the contrary.

Nigel: +1

Pierre: I can propose it.

RESOLUTION: Mixed units are permitted, Pierre to propose a removal of the possibly misleading "either".

github-bot: end topic

IMSC Tests

Pierre: Not much to discuss, should we arbitrarily say that the tests are complete at some
... time, to put a line in the sand? I feel that we're probably there so we could just declare
... version 1.0 open, and the next big milestone is to say when IMSC 1.0.1 is published,
... have a v1.0.1 or whatever.

Nigel: Don't we need to add the 1.0.1 tests in?

Pierre: My plan was to wait until CR at the earliest, based on implementations.

Nigel: Right, because the imsc-tests suite is for Recs and the CR implementation report
... is based on tests that are somewhere else?

Pierre: Exactly, that's how it is in my mind.

Nigel: Makes sense to me.
... I suppose one option would be to create a branch with a mirror of the CR tests so its
... ready to pull request in when the spec moves on.
... It doesn't matter when you do that, and it could potentially cause confusion doing it too soon.

Pierre: Exactly.
... Should we declare this as v1? What do other groups do?

Thierry: I don't know the answer.

Nigel: What does this mean? Add a release tag, put it on the implementations wiki page and maybe the TTWG homepage?

Pierre: Exactly, and send a message to the public reflector.

Nigel: Is there a test harness or framework that can put the input documents in and
... compare the outputs?

Pierre: On imsc.js there's a PNG comparison script, which I'd happily contribute.

<pal> https://github.com/sandflow/imscJS/blob/master/script/refpngcompare.py

Nigel: For me the key thing is that the tests are there - having a test runner would be
... great but could be a v2 thing perhaps.

Pierre: Exactly, the key thing is the tests are stable.
... The PNGs are generated by imsc.js plus I've checked manually that the output is
... correct.
... I found many bugs in the TTML1 tests that were adapted. It's possible there are still
... bugs, but I've seen a definite slow-down in bug reports for IMSC.js and IMSC1 so
... declaring 1.0 is not crazy.

Nigel: Works for me!
... I guess we should feed some of the fixes back into the TTML1 suite too.

Pierre: Yes, unfortunately I haven't kept track of them.

Declare a v1 release

WebVTT review feedback

Thierry: About a month ago David came to one of these calls requesting that the TTWG
... review the latest WebVTT draft and he gave a 2 month period in order to do so. So it's
... now about a month in, and David said he didn't receive any comments, so the CG would
... like to move forward and publish a new wide review draft - we had one already back in
... 2014 but since then there were a lot of new features added to the spec, therefore we
... are going to trigger a second wide review and horizontal review. David considered that
... the comments that were sent during the first review are considered ok.

Nigel: Are we reviewing the draft or the disposition of comments?

Thierry: The disposition of comments and the edits that they triggered to make the
... recent draft. Now there is going to be another review for the new features that have
... not had a chance for wide review except by the CG.

Nigel: Did you say he gave 2 months and now he wants to cut it to 1 month?

Thierry: No he said a month. The TTWG has not really reviewed the answers to the
... commenters' comments, which is what we required.

Nigel: Right so there was a draft disposition of comments and a request for approval by
... the WG and so far there has been silence?

Thierry: Yes, there is no real disposition of comments, it's more issues raised in the
... tracker and in github but there's no real disposition of comments as done for our specs
... for example.

Nigel: I thought you made a wiki page?

Thierry: Yes but it's not finished in my view.

Pierre: Can't we just use the same process? Isn't it a W3C process?

Thierry: The Process doesn't describe exactly what the disposition of comments should look like.

Pierre: What does the Process say?

Thierry: That you have to go through a wide review and that you have to demonstrate to
... the Director that the comments have all been responded to and if the commenter is
... satisfied.

Pierre: I think we should use the same process we have used before.

Thierry: I would agree.
... For example we can not track if the commenter is satisfied or not at the moment.
... If the co-chair believes that what we have now is good enough, even though it is different
... from the process we have successfully followed for years including for IMSC and TTML
... then why not try it? There's nothing that says what the disposition of comments looks like.

Nigel: But we can decide?

Thierry: I would like to decide with David present.

Nigel: It sounds like the response we need to give to David is as follows:
... * We need a clearly set out list of comments and resolutions, with the response from the commenter for each.

Thierry: I think that's a good way forward.

Pierre: My point is just that we should not reinvent the process each time.

Nigel: Thierry please could you go back to David and explain that it's hard for the WG to
... review the comments without a clearly set out list?

Thierry: Yes. Also we have time to do this if there's another WD to incorporate all the
... comments previously received and for the new WD, and we should have time to review
... those during TPAC. However David wants to go to CR before TPAC.

Nigel: Is there a proposed end date for the WD review period?

Thierry: David wanted the review to end in August and I said it's unlikely that we can do that
... because first we need another WD mentioning that its a wide review with the end date,
... so we have to publish that new document, then I have to send emails to groups
... requesting review, then we should have at least 4 weeks of review because its summer.
... For IMSC 1.0.1 we had a month and a half, for TTML2 we had 2 months, so for this one
... we should have at least 4 weeks, which is the recommendation (not a must) in the W3C
... Process.

Nigel: So even if they published today it would be the beginning of September at the
... earliest before the review end period.

Thierry: Yes, however David is away at the moment. It could be that the Editors can
... publish with the automatic tool or I could do it but I need to have a review end date at
... least. Either we choose it today or we propose something to David or we wait for David
... to return.

Nigel: I would propose at least 6 weeks given that its summer, but it would make sense
... to have David's view.

Thierry: Can we agree on it? For me its the minimum given the new features added to that
... spec. Silvia sent a list of the changes.

Nigel: Is that on public-tt?

Thierry: No, I can forward it. I asked her to make a list of the changes like we did for TTML2
... so we can easily refer to the list to define what needs review.
... She has already done that list.

Nigel: Is there a dependency on us reviewing the existing disposition of comments before
... publishing a new WD?

Thierry: I don't think so, we need it to show the Director before moving to CR, and the
... disposition will be a compilation of the comments from the first and second reviews.

Nigel: OK it seems that we have a consensus on 6 weeks based on those present today,
... as the WD review period.

<scribe> ACTION: tmichel to write to David and the CG saying we should publish a new WD with a 6 week review period. [recorded in http://www.w3.org/2017/08/03-tt-minutes.html#action02]

<trackbot> Created ACTION-501 - Write to david and the cg saying we should publish a new wd with a 6 week review period. [on Thierry Michel - due 2017-08-10].

<scribe> ACTION: tmichel Explain to David that we need a more consistent Disposition of Comments for WebVTT. [recorded in http://www.w3.org/2017/08/03-tt-minutes.html#action03]

<trackbot> Created ACTION-502 - Explain to david that we need a more consistent disposition of comments for webvtt. [on Thierry Michel - due 2017-08-10].

TTML2

Nigel: I just want to mention that I joined the Privacy IG's meeting on Thursday last week
... to discuss TTML2 and my notes are at:

TTML2 Horizontal Review - minutes from privacy WG today

Nigel: I don't think anyone in the meeting thought that any of the points raised were
... particularly concerning but did think they are worth mentioning.

Future meetings

Nigel: Next week I've scheduled our meeting for 1 hour, then I'm away for 2 weeks, so unless
... anyone wants to step in to chair then the default will be no meetings on 17th and 24th
... August, returning on 31st August.
... Okay, we're out of agenda for today, let's adjourn. Thanks everyone!