See also: IRC log
<scribe> scribe: nigel
Nigel: For today we need to cover
publication timelines, and there are some TTML2 issues
... marked for the agenda.
... Any particular points anyone wants to make sure we cover,
or Other Business?
group: [silence]
Nigel: Nobody has come back to me
objecting to moving the basis of the meeting time
... to UTC from Boston time, so that's what I'll do as of next
week, assuming no
... objections in the intervening period.
Pierre: By UTC you mean ZULU or UK time?
Nigel: UTC time
Pierre: That will shift the meeting time during DST - why do that?
Andreas: I'm not sure if you were
on the call last week when I proposed this. It was
... discussed at the AC meeting, and lots of members said
Boston time is not very
... convenient to see when a specific meeting is, so they
proposed to shift to UTC
... instead of Boston. Although I think it is not an official
rule there was overwhelming
... support for it and some groups already shifted to it.
... I just proposed it last week.
Glenn: When will it change our time, at the transition of DST, right?
Nigel: Yes
Glenn: Other than 2 days a year it will make no difference?
Nigel: We would be able to choose when to shift to track DST.
Glenn: I don't see any problem with that.
Nigel: We've just requested
transition of IMSC 1.1 to CR2.
... Looking ahead we're about to request transition of TTML1
3rd Ed to CR2,
... and we've just transitioned TTML2 to CR2.
... However some issues have been raised on TTML2 since CR2
that can most
... straightforwardly (in terms of the Process) be handled by
issuing TTML2 CR3.
... There was concern about pushing back the Rec date of TTML2,
which this would do,
... but the flip side as Glenn noted in the week is it will
give more time for implementation
... and testing work.
... Given that there are some changes being proposed to TTML2
that look important
... and that are probably substantive, even though I could try
making the case that they
... are editorial, and given where we are with TTML1 3rd Ed, I
think I'd like to propose that
... we move to a mode where we publish TTML1 3rd Ed and TTML2
transitions
... simultaneously, i.e. TTML1 3rd Ed CR2 and TTML2 CR3, and
then later PR and hopefully
... Rec, and possibly even IMSC 1.1 Rec at the same time if
that is possible.
... Any thoughts on that proposal?
Glenn: As I think I pointed out,
this puts a 47 day delay into the TTML2 schedule as it
... currently stands if we don't try to track TTML1 3rd Ed and
if we don't make that change
... at least considering it substantive. Instead of Sep 13 for
Rec it changes to Oct 30.
... The advantage of putting them in sync is that they would
all go to PR at the same date
... which is Sep 25 to make an Oct 30 final Rec date and to do
that we would need to
... go to CR3 on Aug 9, which means we would have to start the
CfC in time to do that,
... which by my estimation would have to begin July 26 at the
latest or possibly July 25.
... That gives us only a few days to prepare CR3 which I think
is doable if we limit the
... scope of what we are doing to just a few pull
requests.
... Some of those pull requests on TTML2 are marked for the PR
milestone. We could
... do them during CR3 or at PR as long as they are
editorial.
... We could get a new CR3 draft out the door for a CfC then
publish on Aug 9 on the
... same day as TTML1 3rd Ed CR2. I think it's feasible. I'd
like to hear from anyone if
... there are any views about that.
Cyril: The situation is not
ideal; TTML2 has already been delayed for too long so we
... should try to minimise any delay if we can. Especially if
the delays are created by the
... Process we should check with W3S if there are any ways we
can mitigate that, like
... changing the reference on the way to PR. However we aren't
ready for PR for TTML2
... anyway because we don't have the implementation report. So
I'd rather make sure
... we go to PR for TTML2 because we have found a date that is
reasonable in terms of
... producing the implementation report and the test suite,
which is really the problem.
... The deadlines for publication are internal to W3C and W3C
should be able to resolve
... that problem. The situation is becoming critical because we
have a chain of dependencies.
... Not just TTML2 on TTML1, but also MPEG, which needs IMSC
1.1 to become finalised.
Nigel: What's the deadline for that?
Cyril: I will check.
Pierre: Last time I checked Oct
30 was an appropriate deadline for MPEG.
... I really like the idea of making it simple on ourselves and
targeting publication of
... IMSC 1.1, TTML1 3rd Ed and TTML2 all synchronised on Oct
30.
Cyril: I don't think that makes
it simpler for us. The sooner we can get a spec out of
the
... door the more time we have to focus on the next one.
Pierre: The last month is no effort for us, so I don't think it makes a big difference.
Glenn: The time between the PR
and the Rec has some stretch. Right now with these
... dates we would have to go to PR on Sep 25 for Oct 30 Rec,
which is probably longer
... than is needed, but that is what Philippe's schedule tool
produces right now. The
... upside is that instead of Aug 9 for TTML2 PR which is
coming very quickly we would
... have until Sep 25 which works out to I think 47 days extra
for work on test suite and
... implementation report which would be a very useful
additional space in which to do that.
... Completing them by Aug 9 is going to be very
challenging.
Nigel: +1
... I don't see how it would even be possible given where we
are, realistically.
Cyril: What is a reasonable date
for producing the implementation report and test suite?
... Regardless of a CR3 when is the earliest we could produce a
PR?
Glenn: We are only one party to
the implementation report. I'm having to timeslice
between
... spec work and test suite writing. I'm nominally on holiday
right now until 23rd July,
... and will be back on 24th, so that would technically leave
me 16 days...
Nigel: The question is how much time do we really need to finish the test report?
Glenn: I can't answer that question.
Cyril: So we are not near PR?
Glenn: I can't say that right
now.
... I have pushed most of the validation tests to the test
suite but not any of the rendering
... tests yet. My expectation is that other implementers will
contribute tests as well.
Cyril: We should, yes.
Nigel: Yes.
Andreas: Is there an overview of
which in-flight tests are missing?
... It is possibly on all of us to contribute something so if
we could coordinate this then
... it could speed up the process.
Nigel: +1
Current draft Implementation report spreadsheet
Cyril: [has to drop off the call]
Glenn: That spreadsheet does not
show mapping between features and tests yet. It is on
... my work list to add a table that does that and I probably
won't be able to start work
... on that until 26th July given I will have to focus on CR3
completion before then.
Nigel: Thank you
Glenn: This activity is part of Skynav's internal implementation activity.
Nigel: It's a tricky thing to coordinate contribution of tests - I'm happy to take any suggestions!
Andreas: For the specifications
we have an Editor. For the Test Suite I don't know if
anyone
... has the lead on it. If someone like Glenn is actually
working on some tests, say there
... are 50 missing and Glenn is working on 25 then it would be
inefficient if others would
... also work on those same 25. Just my immediate thought.
Glenn: The way I've organised the
current TTML2 tests repo is I have 3 directories:
... valid, invalid and resources, which test for validity,
invalidity and resource files.
... My plan was to add a presentation directory as well for
populating presentation tests.
... We could have, say, a contrib directory under which
organisations that wish to contribute
... tests can be done, then tests can be merged from
contrib/XXX into the top level as
... we proceed. Under the contrib tests we could allow push
with little oversight but
... we would scrutinise them before moving them to the official
tests.
Andreas: I think that's not I
asked for - it possibly makes it more complicated. If it
is
... clear how the tests are organised then we or I can make a
pull request and make the test
... What I'm asking for is if there are features that need
tests. If there's a list then we can
... see who is working on it.
... The minimum required is, for each feature, at least one
test file and one rendering?
... Or a valid and an invalid and a rendering?
Nigel: I think it's a valid one,
an invalid one and a rendering.
... The Valid ones can be Rendering ones too.
Glenn: True, but the ones written
right now are designed to test the validation process.
... For example many of the style validation tests simply put a
style attribute in a style
... in the head without referencing it. It just tests for
parsing.
... One thing we should keep in mind is we are not testing for
interoperability, but
... to validate the specification, to demonstrate that there
are implementations of features
... and that they are implementable.
... This directs how we create the test suite.
... Having said that, that means that for testing
implementability we don't have to go
... as far as one would typically do in a product for testing
interoperability.
... We don't have a document requirements set for any test
suite in TTWG at the moment.
... Any individual contributing tests up to now has used their
own philosophical requirements.
... I think we should try to do the minimum that satisfies the
PR transition request.
Andreas: I agree.
Nigel: We do effectively have a
list, having all the features introduced in TTML2 and for
... each one those three things, a validating test, a
non-validating test and a rendering test.
Glenn: I plan to work on creating that list so we can populate it with tests we have.
Andreas: This list makes sense.
Then we can see where we can contribute tests.
... For example if there is no TTPE support and no IMSC.js
support then it is difficult to
... get a rendering. I would like to see if we could contribute
something first of all.
... We are only working on features selected by IMSC 1.1. If
Glenn is already doing it then
... we don't have to, but if there are gaps then we can
consider what to fill in.
Nigel: I'm hearing that there are no presentation tests yet.
Glenn: Actually we have a large
number of tests for TTPE at the moment.
... There may be multiple entries for each feature, with a test
name, then validity tests, invalidity tests and presentation
tests.
... Then for each feature you would be able to see the
collection of tests.
... I'll be able to populate that table at least starting with
those 100 test files that cover
... multiple features. Those cover probably most if not all of
the features that are going
... to be in IMSC 1.1. There may be a few exceptions. Since
those are the features we were
... focusing on implementing first we probably have most of
them covered.
... The ones we're focusing on at the moment are those marked
as at risk, aside from the
... audio features, so I'd look to you Nigel to fill that
gap.
Nigel: Ok I can do that.
Glenn: I'm fairly confident that
all of the features will have tests and presentation
support
... for all three categories.
Andreas: Okay that sounds good.
If you can provide the table by next week's call then
... we can start there.
Glenn: I think I can have it in place by next Thursday.
<tmichel> Sorry I must leave for a few minutes.
Nigel: Going back to the top, it
seems that issuing a TTML2 CR3 CfC as early as possible
... next week has general approval and is unlikely to
materially affect the Rec date
... significantly, and there's support for synchronous
publication of TTML1 3rd Ed, TTML2
... and IMSC 1.1 as Rec by end October at the latest. So that's
the plan I'll try to go with.
... In terms of testing, I don't think there's much more to add
- summarising, work is in
... progress, and I'm on the hook for audio related
tests.
... And Glenn will summarise the state of existing tests in a
table by next week's call.
... Any other points before we move on?
Pierre: Since we have agreement
on that it would be good if the Chair could craft the
... timeline, which would be close to what Thierry already did,
and send it to W3 Staff as
... a heads-up so they can highlight any reason it might not
work, and we can know up
... front if we are going to run into trouble.
Nigel: Good suggestion, I will take that action.
<scribe> ACTION: Nigel Draft a timeline for TTML1, TTML2 and IMSC 1.1 publication and share with plh and tmichel as advance notice [recorded in https://www.w3.org/2018/07/19-tt-irc]
<trackbot> Created ACTION-514 - Draft a timeline for ttml1, ttml2 and imsc 1.1 publication and share with plh and tmichel as advance notice [on Nigel Megitt - due 2018-07-26].
Nigel: First thing to note is that the "crafty plan" regarding changing the reference to
<glenn> https://w3c.github.io/spec-releases/milestones/?rec=2018-10-30&noFPWD=true
<glenn> here is PLH's schedule tool working back from REC on 10/30
Nigel: TTML1 3rd Ed from TTML2 is
no longer needed and we are no longer concerned
... about the exclusion period for TTML1 being longer than
TTML2 if we go ahead with
... the plan just discussed.
... Secondly, I did issue the CfC to publish TTML1 3rd Ed CR2
on Tuesday.
... Is there anything else to discuss on TTML1?
Pierre: I think we're good unless
we want to reflect changes in TTML2 back in TTML1!
... We'll have to keep careful track of any substantive changes
to TTML2.
Glenn: There's only 7 pull
requests on TTML2 CR3. Three are the substantive ones
which
... are the change to the normative ref to TTML1, the addition
of image and audio to the
... list of content elements, and scaling back animatability.
Those are my current limited
... set and I don't anticipate any further ones unless someone
pulls something out of the
... hat in the next few days.
Pierre: I remember some modifications to style resolution that I have not studied yet.
Glenn: Right now I think they are
completely editorial. Please do review them. I marked
them
... as PR milestone. We have until Nov 25 or a bit before to
put in place other editorial
... fixes. If someone thinks they are substantive we can scale
them back to make sure
... they are editorial. It is not my intent that any of those
should have a substantive effect.
Pierre: As we merge those things
if they impact TTML1 we should consider back-porting
... them, without prejudice one way or the other.
Glenn: Right, the three substantive ones are not relevant to TTML1.
Nigel: Are there any things to discuss other than stuff marked for the Agenda?
<glenn> https://github.com/w3c/ttml2/labels/agenda
Nigel: That has 8 things on it.
github: https://github.com/w3c/ttml2/issues/443
Glenn: This is the one where we need to send a message to Glenn Goldstein.
Nigel: Yes that's for me. We
began working on this a couple of weeks ago and then
... Glenn came into the discussion - it wasn't at the forefront
of my mind that he is actually
... a member of this WG. So he's seen something in preparation,
but I haven't yet formally
... sent it.
... Action still with me.
... Is there anything to comment on regarding the draft
text?
group: [silence]
Glenn: I would say go ahead and send it, so the issue can be closed.
SUMMARY: Nigel to review status of the draft and send if no further changes are needed.
github: https://github.com/w3c/ttml2/issues/881
Glenn: I put this on the agenda
because I posted a pull request on it, and I had a couple
... of open questions for the group.
... The pull request is #914.
... It turns out that unicodeBidi is discretely animatable in
TTML1. The most we can do
... is deprecate it, and if we do then we should deprecate
animation of direction too. I've
... put that deprecation into this pull request.
... I did three more changes:
... 1. Scaled back continuous animation to only values that
represent alpha, RGB colors, luminanceGain and position
coordinates.
... All other styles that were previously marked as
continuously animatable I have moved back
... to discrete only.
Nigel: I need to ask for a change
there, because the pan and gain attributes need to be
... continuously animatable to be useful.
Glenn: I'm sorry, I did leave pan
and gain. I changed pitch to discrete only, because I
... didn't see a use case for continuous animation of
pitch.
Nigel: I don't think animation of pitch and rate has any use case at all.
<tmichel> sorry I am back
Nigel: Why would you discretely
animate the rate, for example, as opposed to splitting
the
... text up into different spans and putting different rates on
each one?
Glenn: I see what you mean, the
timing of those set transitions related to the text is
... unknown to the author. I'll mark them as "none" for
animation.
Nigel: Sorry I interrupted your list.
Glenn: 2. Add a note under the
section that defines the style properties that describes
... the style property definition tables and to provide some
information on the terms we
... are using in the animatable row, in particular we never
said what discrete and continuous
... means.
SUMMARY: @skynavga to modify speak and pitch to make them non-animatable.
Nigel: I don't see the need to deprecate discrete animation of unicodeBidi and direction.
Glenn: It is more consistent with the CSS recommendation.
Pierre: I don't see a need to
deprecate use of set with a style property just because
it
... doesn't seem useful to use. I think I'm of the same mind as
Nigel that there's no need
... to deprecate it, but I think Glenn is pointing to a bigger
issue?
Glenn: You cannot animate the
content of text but if you [scribe failure]
... For example in TTPE the style attributes are mapped to the
UnicodeBidi control characters
... and if you say they can be animated now you have to re-run
the Unicode Bidi algorithm.
Pierre: Right, but as Nigel
pointed out, set is no different than creating a bunch of
ISDs
... each one with a different value of the style properties. So
from an implementation
... perspective there's no difference.
Glenn: I agree, and I think it
actually works in TTPE that way, but I do want to mention
... the reason this came up was to satisfy the CSS WG
decision.
Nigel: I think that CSS WG decision was about CSS animation including continuous animation.
Glenn: They just wanted to make it not animatable at all.
Nigel: I don't think they make the distinction between continuous and discrete.
Pierre: My feeling on this is we should not spend time deprecating this.
Glenn: Okay, it's easy enough to revert that. I'm happy to deprecate it.
Pierre: Once you've committed to supporting set it really doesn't matter.
Glenn: Yes, that's correct, but it does have a questionable impact on authoring.
Nigel: There's a majority view not to deprecate and no objection to not deprecating.
RESOLUTION: Do not deprecate discrete animation.
github: https://github.com/w3c/ttml2/pull/890
Glenn: Here we have a couple of points from Andreas that I'm not agreeable with.
Andreas: I think we can limit the
discussion on one or two points, the others I can accept as
is.
... My major point is with the reference. It's good to
reference the OpenType specification
... there is an ISO document but that is not accessible because
it has to be paid for.
... But there is a Microsoft version that is also referenced by
CSS and other experts have
... also referred to it in discussions I have. Either reference
it or at least say for the reference
... to the ISO document that the Microsoft one is also
available.
Glenn: It is accessible but it
costs money. We also have a precedent in this group and
others
... that ISO/ITU standards take precedent over national
standards, and those in turn over
... company documents. The Microsoft document is at the bottom
of that list. In this group
... we have always preferred international standards. Wikipedia
states that the OpenType
... standard is the ISO standard and quotes Microsoft as saying
they are synchronised.
... Readers will probably already know about Microsoft's
documents. At most I'd be willing
... to add an informative reference to the MS doc in addition
to the current ISO reference.
Andreas: I'm happy with doing that.
Glenn: I'll add an informative reference to the MS doc and leave the ISO doc as the main ref.
Andreas: One other thing is an editorial matter which you said had a typo. Would you add that text?
Glenn: The comma was already present.
Andreas: Okay
Glenn: I'll consider the proposal.
SUMMARY: @skynavga to consider editorial improvement and add informative reference to MS OpenType spec.
github: https://github.com/w3c/ttml2/pull/897
Glenn: Added to agenda to prompt for review.
Nigel: I'm mid-review. One
technical question: Glenn you added these into
... the audio.attlist attributeGroup, whereas I saw an
alternative way which is to add it
... to the Audio.content.attrib.class attributeGroup.
Glenn: Simple explanation - it's a global not a local attribute without a namespace.
Nigel: Ah, interesting.
Glenn: Our convention is that global attributes are put into classes.
Nigel: Okay, thank you!
SUMMARY: Review to continue
github: https://github.com/w3c/ttml2/pull/898
Glenn: I think I just wanted to
note that I've moved this to substantive. We discussed
earlier.
... Pierre, you might want to take a look at this. It adds
audio and image to content element.
... During the last meeting you had some questions and were
reluctant to make this change.
Pierre: Alright, em...
Glenn: I went through all
instances of the phrase content element in TTML2 and
found
... no cases where it was not appropriate.
Pierre: I sense that we've
already had this discussion somehow and you argued for
the
... opposite in the past, but you've done the research now?
Glenn: Yes. It took me about a day!
Pierre: I need time to think
about it. I'll study it.
... You're saying it's not embedded content, just content?
Glenn: It's still covered under
embedded but also in content element. We already had
... "presentation related" which included image but not audio.
This attempts to make it
... consistent.
Pierre: I'll add this to my review list, thanks for pointing me to it explicitly.
github: https://github.com/w3c/ttml2/pull/899
Glenn: This is a small number of
editorial improvements.
... Nigel you have an outstanding comment I need to look
at.
Nigel: Yes
Glenn: We can do this offline.
Nigel: Ok
github: https://github.com/w3c/ttml2/pull/911
Nigel: Glenn, you and I had a bit of back and forth on this one.
Glenn: Yes, I need to process it
some more, and I don't think we can resolve it right now.
... It is marked for the PR as editorial so we have some time
to get straight on this.
Nigel: Okay, let's continue offline.
Nigel: I'm not sure we have
anything to cover on this today?
... Thierry is there any news on the transition?
Thierry: No, I sent the TR on
Monday and still don't have any news. Plh is on vacation this
week
... so I don't know if Ralph will process the transition
request tomorrow. Maybe on Monday.
... We have until Wednesday morning for me to get the approval
to make the publication
... request for publication on Thursday next week so we still
have time.
Nigel: Thank you for the update.
Pierre: Nigel you opened an issue, but I think it was answered.
github: https://github.com/w3c/imsc-vnext-reqs/issues/36
Nigel: I think we're out of time
to do anything about this in IMSC 1.1. We didn't action
... this in the CR2 transition request.
Pierre: No, and the idea really
is to modify the Image profile as little as possible, if at
all.
... I've not heard folk being unsatisfied with Image Profile.
Users of the Text Profile were
... quite vocal with strong opinions, but I've not seen any
documented feedback on the
... Image Profile so my philosophy is not to modify it without
anyone expressing that opinion.
<atai2> +1
Pierre: Same applies to position.
Glenn: Did you add the new image element?
Pierre: Yes, and deprecated smpte:backgroundImage.
Nigel: I take your point, though it does seem inconsistent.
Glenn: How about tts:showBackground? Everyone seems to want to change that.
Pierre: If you're using an image
you'd typically use a transparent background.
... Basically I've seen a lot of full raster images for Image
Profile.
... What we can do about this is either close it as "won't do"
(for now), or just leave it
... open and mark for v.next. I think I'd rather just close
it.
... Here the question is if someone cares. If nobody cares we
should close it. I don't care!
Nigel: I don't have any plan to
use Image Profile for anything, so it makes no difference
to
... me in usage.
Pierre: My suggestion would be to close it.
SUMMARY: No immediate requirement to `#initial` has been identified, closing with no action until such a requirement is raised.
Nigel: I've labelled it as Wontfix and closed it.
Nigel: I need to add TTML Profile
Registry to agenda for next time, since there's been
... some discussion on it.
... A reminder that Thierry sent details of early bird discount
period for TPAC, so if you have
... not registered but plan to, do it soon and save some
money!
... Unless someone wants to step in and Chair, we may not have
meetings on 16th and 23rd August while I'm away.
... If you do want to Chair, please let me know!
Pierre: I'll be away those dates too.
Nigel: Okay, thanks everyone! [adjourns meeting]