- From: Nigel Megitt <nigel.megitt@bbc.co.uk>
- Date: Thu, 20 Apr 2017 17:05:46 +0000
- To: Timed Text Working Group <public-tt@w3.org>
- Message-ID: <D51EA898.3D4A0%nigel.megitt@bbc.co.uk>
Thanks all for attending today's TTWG meeting. Minutes can be found in HTML format at https://www.w3.org/2017/04/20-tt-minutes.html
In text format:
[1]W3C
[1] http://www.w3.org/
Timed Text Working Group Teleconference
20 Apr 2017
See also: [2]IRC log
[2] http://www.w3.org/2017/04/20-tt-irc
Attendees
Present
Philippe, Mike, Pierre, Nigel, Glenn, Andreas,
David_Ronca
Regrets
Chair
Nigel
Scribe
nigel
Contents
* [3]Topics
1. [4]This meeting
2. [5]TPAC 2017 Advanced Planning
3. [6]TTML (process)
4. [7]TTML (Prioritisation and assessment relative to
planned timeline)
5. [8]TTML (issues)
* [9]Summary of Action Items
* [10]Summary of Resolutions
__________________________________________________________
<scribe> scribe: nigel
This meeting
Philippe: I will have to leave before the bottom of the hour.
Nigel: For today we have TPAC planning, TTML, IMSC and a new
agenda item for HDR in PNG.
... I have one extra issue for TTML, which is the inline boxes
one, as discussed on the reflector.
Mike: I'd like to spend a minute or two on how the normative
references are interpreted
... when there is no assigned version.
Andreas: I have an additional proposal, to look at the timeline
we made in London
... regarding TTML2 and IMSC 2 work and see where we are. I
think that also affects TPAC.
Nigel: Thanks, any more points to raise?
group: no more
TPAC 2017 Advanced Planning
Nigel: Does anyone currently have a preference for meeting on
any of the days particularly?
Andreas: I have no preferences but would like to avoid
conflicts with other meetings I want to attend.
Nigel: Any groups that we want to avoid clashes with? Or hold
joint meetings with?
Andreas: The last 2 times we joined the Web & TV IG. In 2016 it
was official, in 2015 it was
... a Chair's report. I assume it would be useful to try not to
clash with that group.
Nigel: Agreed.
Andreas: We have also discussed CSS related issues so we should
think about requesting
... a joint meeting.
Nigel: Yes let us certainly do that, with the scope being to
present styling requirements
... for subtitles and captions that cannot currently be met by
CSS.
... OK I will complete the form asking for any 2 days and
avoiding the groups we just mentioned.
... If anyone has other points to raise on this, or
constraints, please let me know.
... The other thing I need to mention is that if anyone has
difficulty attending the meeting
... in person, perhaps because of travel difficulties, please
let me know in case W3C can
... assist e.g. with remote attendance.
TTML (process)
Nigel: Last week we discussed the process, and the Chair was
urged to discuss it with the Editor.
... That happened, and other conversations also. As a result I
would like to reiterate that
... we have a process that we agreed previously,
[11]https://github.com/w3c/ttml2/blob/gh-pages/EDITING.md#ttml2
-editing-guidelines
[11] https://github.com/w3c/ttml2/blob/gh-pages/EDITING.md#ttml2-editing-guidelines
Nigel: And we should stick to that. A reminder that as a normal
rule aside from minor
... editorial changes pull requests will be left open for a
minimum of 3 days before merging
... and if anyone wishes to request extra review time then that
will normally be granted up
... to 14 days.
... As Chair I also would ask that everyone is as flexible as
possible in allowing the Group
... to make progress given that we are past our London-agreed
deadline already.
Pierre: What is the policy on pull requests when somebody
submits one, in terms of best practice?
Nigel: First of all the Editor has requested a branch naming
convention, which is a convenience, but
... for me not a primary requirement.
... Secondly if a totally different proposal for an issue is
made then it should be in a different
... branch with a different Pull Request. However if
modifications need to be made on a
... Pull Request they should be made on that request's source
branch.
Philippe: Sorry I can't stay longer. Bye everyone [leaves]
TTML (Prioritisation and assessment relative to planned timeline)
Nigel: In the agenda I noted that we have passed our
self-imposed deadline for publishing
... a WD for WR and that there may be a need to review the
priorities of issues and possible
... drop the axe on some of them now. Andreas, you wanted to
raise this also?
Andreas: In London we set a positive schedule, which we have
missed, but I do not know
... where we stand and how far away are we from what we
originally planned, and then
... what is the new realistic deadline and how it can be
acheived?
Nigel: OK, that's an important question, thank you.
Pierre: Before we consider deferring issues we should think
about the best process for
... disposing of issues going forward. We should talk about the
process for going forward
... with TTML2. In London we discussed a process, where we went
through TTML1 issues
... and marked them as "discussed and agreed" and those have
not yet been fixed.
Nigel: You're asking about the editorial progress in
implementing those issues?
Pierre: Yes.
Glenn: As a blanket statement that is not correct. Some of the
discussed and agreed issues
... have been closed, others are effectively done and marked as
To close? which I would
... have closed if we did not wish to wait for consensus before
closing an issue.
... There are a couple of discussed and agreed issues that have
had work done but not been
... updated. So progress is being made, but it is not complete.
... The list I am working from is the open issues under the
TTML2 WR milestone, which
... is showing 58 open, of which 4 are marked as "to close?" so
that could be turned into
... 54 pretty soon.
Nigel: And where are we up to with those 54?
Glenn: I've been prioritising the Discussed and agreed first.
There are some that come from
... TTML1 changes that are not raised yet, for example #221 on
TTML1 and Andreas's
... pull request which I processed further last night. So there
are a couple of TTML1 issues
... that we need to bring in. Otherwise I am making my way
through this. There are quite
... a few that have had some work done that need to be wrapped
up, for example around 5
... issues that pertain to the new annex defining semantics of
display extent, which we
... talked about quite a bit in London, so those can be dealt
with by wrapping up that work.
... I'd say that most of these 54 have had a fair amount of
discussion and some may
... require more technical discussion in the group than others.
Maybe a third should be
... fairly straightforward to wrap up.
Nigel: That would be around 18 being 54/3.
... Given that there's a fair amount of effort to do, the
obvious thing to me would be
... to bring in more editorial effort by other group members
than the Editor to draft
... pull requests to resolve issues.
Andreas: I understand it is important for some people to reach
Recommendation by the end
... of the year. One way to handle this with so many open
issues is to mark some issues
... as being less high priority so the TTML2 spec would still
be valid if we did not address them.
Pierre: Going down one path you hinted at, I've heard folks
from Netflix being really keen
... on completing this work, and Movielabs, and the BBC and
perhaps others. I'd be
... surprised if we couldn't find resources there to help work
on editing the document.
... That's a path that perhaps we should explore.
Glenn: There are two issues to separate: one is generating
draft text to incorporate, and
... the second is doing final editing on that if necessary. We
need the former more, which
... is to generate pull requests to resolve particular issues.
I have said on multiple occasions
... that I welcome other members generating pull requests to be
contributed for inclusion.
... If other resources are available that's what I'd like to
see done.
Pierre: I think it needs to be more than that because if the
Editor is unable to process all those
... pull requests then we would be in a similar situation.
Nigel: To be clear, my comment earlier was about one individual
both drafting and merging
... rather than just merging. However it certainly is common
practice in W3C to have
... multiple Editors for a specification and I would encourage
that, and Philippe has indicated
... to me that he would support that. It removes the
opportunity for bottlenecks. However
... I think Glenn has mentioned before that he would like to
retain a single view for
... consistency.
Pierre: If this is something that the group wants then the
Chair could ask proponents for
... more editing help.
Glenn: I should mention that I have cleared my schedule in May
to deal with issues.
Pierre: I would recommend that the Chair assigns issues to
specific people to dispose of,
... and separately I recommend raising the issue with
proponents regarding resource
... availability for providing more Editing.
Nigel: It is not clear to me why the Chair needs to assign
issues - if proponents are
... interested in taking on an issue then I suggest they do so,
as a first step add a comment
... to the issue stating the intention to resolve. In general
in W3C we work on the natural selection model where things that
people are interested in get resolved. Assigning issues
proactively will prevent other interested parties from
contributing, so I am not in favour of doing that.
Pierre: There are some open pull requests, for example one on
TTML1 raised in February.
Andreas: The TTML1 one was requested in London within two weeks
otherwise Glenn would
... propose something else. I tried to fix it as soon as
possible but then it seemed it was
... not needed that urgently. It's not a big issue for me. In
terms of getting more people
... involved it would be more motivating for people if we deal
with pull requests in time.
Glenn: Thank you for providing that input very quickly. It was
then waiting for me to do
... final editing. So in terms of final editing I have been
giving it some thought and finally
... made the changes I thought were necessary yesterday. There
are also some comments
... on an earlier commit.
... One of the issues I have is that issues seem to need full
group discussion before closing.
Nigel: In fact that can be done on the issue.
... It does not need the group to discuss it in a meeting.
... By the way we do not have a group policy for how long pull
requests should take to
... process, but it seems reasonable to me that it should be
days or a small number of weeks
... and that "months" is too long.
David_Ronca: Looking at the WR, there are comments that first
it was March, and now it
... looks like it will be June, and I don't see any strong way
that we can ensure we will
... publish the WR in June. We are beginning to get the
perception that we will never get
... there. We have a lot of time and money invested in this
project, and nobody can move
... until we make progress. I don't feel like we're leaving
this meeting today having made
... progress.
Nigel: What progress are you hoping for?
David_Ronca: Perhaps some agreement on how we can meet WR in
June.
... I agree that we need to delegate issues out. We need to
start making some hard decisions
... about what's in or what's out. If there's some way that we
can contribute more to try to
... get these resolved then I would.
Nigel: Working backwards that means that we need to have pull
requests open or merged
... for all issues to be met in the WR by the end of May so
that the group has enough time
... to review and get to consensus on publishing those
resolutions as a WD. If there are
... remaining issues open at that point without pull requests
then we would need to make
... hard decisions at that point.
Pierre: Issues raised externally will not go away, they will
still need to be addressed.
Nigel: That's true, however we do have an opportunity to
resolve external issues between
... WD and CR.
Glenn: The disposition can be a deferral.
David_Ronca: I would hate to begin work on an issue and find
someone else is doing it.
Nigel: Well you could assign yourself to it, and also add a
comment so if anyone else is
... working on it then they can mention that.
Glenn: Also I would suggest asking me in case I've already put
some thinking into it.
Pierre: That's inefficient. If you've already begun to think
about it you should assign the issue to yourself.
David_Ronca: I would agree with that - if we can assign
resource to make a sprint on this
... then I would like to see issues assigned, and non-assigned
issues need to be assigned
... to somebody.
... I'm actually willing to try to get my team to make some
contributions to push through
... this. I want to do it efficiently. I would appreciate if
every issue being worked on has an
... assignee. Is that a reasonable request?
Glenn: Yes, they should at least do a first attempt to read the
comments. I will go through
<atai> +1
Glenn: and see if I have given anything significant thought and
not documented then I may
... assign it to myself.
David_Ronca: That would make it much easier because then I can
filter the unassigned issues.
Glenn: I need to do a pass through and review the current
status.
... There's also a label "Request help" so it might be useful
to make more use of that label.
David_Ronca: Glenn please could you assign the ones you are
working on to yourself?
Glenn: Sure, that seems reasonable.
... I'll do it this week or next week.
Nigel: Okay, that seems like good progress, so we have a way to
identify what needs
... working on and when something is already being worked on,
some offers of more
... effort, and some offline discussions to have. Thanks.
David_Ronca: I have to drop off now. [leaves]
Andreas: Me too. [leaves]
TTML (issues)
Glenn: I'd like to raise #292.
[12]Update note regarding ISOBMFF and DASH.
[12] https://github.com/w3c/ttml2/issues/292
Glenn: This is an issue pointed out by Mike on email recently,
with reference to DASH,
... ISO BMFF, CMAF and so forth. I committed this directly
since it is an informative change.
... If that is acceptable I would like to close it.
Mike: It's editorial, not a big deal, thanks.
Nigel: I'm happy with it too.
... I've closed that issue.
... Apologies Mike we haven't been able to get around to your
question about
... unversioned normative references, and now we have no staff
on the call to assist.
Mike: That's okay, it can wait until we have Thierry of
Philippe on the call.
... I noticed that some references have versions or dates and
some don't and wonder how
... we deal with those.
Glenn: For tracking and transparency purposes we should
probably open an issue on this
... post it as a question, adding the "question" label.
Mike: Sure I'll do that.
... It may result in changes of some sort but I'm happy to
start with a question.
Nigel: Can we raise the inline boxes issue?
[13][css-inline] Define the content area of inline boxes
[13] https://github.com/w3c/csswg-drafts/issues/814
Nigel: Am I correct in understanding that the proposal on the
issue is essentially the same
... semantic that we want with a different syntax?
Glenn: They have not yet proposed any semantic. They would also
have to deal with the
... fact that height and width do not currently apply to
non-replaced inline boxes. There's
... nothing in CSS so far that alters that. The second thing is
if there are keywords that would
... satisfy our needs. I think we should not make our work
dependent on their process.
Nigel: My proposal is that we define the semantic that we want
and add it to the CSS issue.
Glenn: We can do that. First we need to agree what we want to
achieve. We need a concrete
... proposal on our side. We have two action items, first as
agreed in London to change
... the semantics of bpd and ipd so it does not trigger the
inline block semantic and second
... to define a way using bpd or a keyword to mean what we want
it to mean. When we have
... that at least in an ED we can provide a reference to that
for the CSS folk.
Nigel: Those are both things you have been working on Glenn -
where are we up to on them?
Glenn: Yes.
[14]Ways to make span background height be lineHeight
[14] https://github.com/w3c/ttml2/issues/150
Glenn: I will assign that to myself and create a new issue to
track the agreement to add
... a new keyword to the display property called inlineBlock
and change the semantics in a
... few places that had automatically triggered that behaviour.
... Since it changes something we've had for a while some
implementations like TTPE will
... need to be changed accordingly.
Nigel: I've assigned #150 to you Glenn. This also affects #146
and #163.
... In my view the separation of inline block semantics is the
resolution to #146.
[15]Inline block semantics impacts text wrapping
[15] https://github.com/w3c/ttml2/issues/146
Glenn: You're right, yes.
Nigel: I see you're already assigned to #146.
... On the point about HDR I see that I have just received an
email from a colleague
... giving me the go-ahead to make a BBC submission to add to
the annex about HDR an
... alternate way to convert sRGB using hybrid-log gamma. I
will do that shortly.
Glenn: Is this a replacement text?
Nigel: No it's an additional method.
[16]Add support for compositing a TTML2 document onto HDR video
[16] https://github.com/w3c/ttml2/issues/233
Pierre: The reason for removing the to close? label was
documented in the minutes.
... I will create a PR to address the current comments.
Nigel: I see that Pierre and I are assigned to this. I will
create a PR for the HLG way of doing it.
Glenn: We have not discussed the name yet in the group. I think
the simpler name is better.
Nigel: Pierre and I have already made clear our preference for
the more specific name.
Glenn: I do not think we will need to use luminance for
anything else.
Mike: Luminance is a term that goes way beyond what we mean
here so I would prefer that
... it be qualified in some manner. The original proposal was a
little verbose but we need
... something more than just luminance.
Glenn: How about luminanceGain?
Pierre: I'm happy with luminanceGain, I don't think you are
Nigel?
Nigel: My test question is: can it be used in any other
capacity than for PQ HDR?
Pierre: It is an absolute gain. [some discussion]
... To get to a PQ encoded image you need absolute luminance,
which you get to with this
... metadata.
Nigel: Is it the only way it can be used?
Pierre: It can be used to create an HDR image for other
encodings than PQ.
... It does not specify anything only specific to PQ. The value
can be used for other purposes.
... The end result of applying this is that you end up with an
XYZ value with absolute
... luminance that you can then do what you want with.
... Fundamentally the issue is that sRGB is limited to 0-80
cd/m. There are schemes that
... can use a broader dynamic range. The goal is to be able to
produce images that can be
... used in any number of schemes, one of which is PQ but
conceivably others.
Nigel: And my understanding is that XYZ is essentially
canonical, so if we define
... luminanceGain to be the thing you use to get to a canonical
format then I'm happy with that.
Pierre: Yes it is the thing that gets you to absolute luminance
values.
Nigel: Ok then let's go with luminanceGain
... If there is some unexpected further problem with that then
we can deal with that later.
Pierre: I will make a PR to deal with that and the other
comments.
Glenn: Ok that's good.
Nigel: I added comment
[17]https://github.com/w3c/ttml2/issues/233#issuecomment-295787
705
... What can we look at next? I think we have done enough in
the offline discussion to
... deal with the smpte discontinuous issue?
[17] https://github.com/w3c/ttml2/issues/233#issuecomment-295787705
[18]No restriction on time expression for smpte discontinuous
mode
[18] https://github.com/w3c/ttml2/issues/161
Glenn: I'll do a PR for that.
Nigel: Thank you!
... I have assigned it to you Glenn.
... Did we get anywhere on whitespace handling?
[19]Ambiguity regarding tab (U+0009) processing in significant
whitespace.
[19] https://github.com/w3c/ttml1/issues/235
Glenn: I sent an email to Steve Zilles because I could not
resolve this by looking at the XSL
... specification. CSS now says that a significant tab gets
mapped to spaces - I would like
... to avoid using those semantics if possible. My preference
is that if a tab is significant
... then either treat it as zero space or exactly one space but
I need other input.
Pierre: The XSL spec is really clear - tabs are preserved but
it does not say what a tab is!
... Where do you see the ambiguity in the XSL spec?
Glenn: Two issues: one is what do the xml:space semantics mean
with a tab? ...
Nigel: I just want to step in here - the explanation is in the
issue. Pierre is there a problem with that?
Pierre: Yes, and I will add that to the issue. The issue is
that CSS changed. To my mind you
... should just not put tabs in TTML documents! It doesn't say
what happens.
... A solution, regardless, is to say authors should not put
tabs in their documents.
Glenn: That's not an adequate solution because they appear all
the time. For xml:space="default"
... we need to define whether there is a case where a tab may
appear or if it is always
... normalised to space. I think I convinced myself that tab is
mapped to space in this case.
Pierre: I'll take an action item to comment on the thread.
Regardless I think tabs should not be present in the document.
Nigel: The question is not if it should be allowed in a
document but to define what to do with
... it if it is present.
Glenn: If it is mapped to space then we do not need to deal
with it, but if it is preserved then
... it falls into the reduced infoset and it needs some
presentation semantics. We have some
... options, to treat it as 0 space, 1 space or like CSS does,
some multiple of spaces.
Nigel: What gave rise to this issue is that if
xml:space="default" some tabs are preserved.
Glenn: Only if they are the first character of an element or
the first character after a new line.
... That was the question I posted, maybe I'll take the text of
my question to Steve and Tony
... and paste it onto the comment so that people can see what I
was asking.
Nigel: We're out of time, thanks everyone. [adjourns meeting]
Summary of Action Items
Summary of Resolutions
[End of minutes]
__________________________________________________________
Minutes formatted by David Booth's [20]scribe.perl version
1.152 ([21]CVS log)
$Date: 2017/04/20 16:59:41 $
[20] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
[21] http://dev.w3.org/cvsweb/2002/scribe/
Received on Thursday, 20 April 2017 17:06:19 UTC