- From: Nigel Megitt <nigel.megitt@bbc.co.uk>
- Date: Thu, 27 Apr 2017 16:29:19 +0000
- To: Timed Text Working Group <public-tt@w3.org>
- Message-ID: <D527DA4F.3DF54%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/2017/04/27-tt-minutes.html
In text format:
[1]W3C
[1] http://www.w3.org/
Timed Text Working Group Teleconference
27 Apr 2017
See also: [2]IRC log
[2] http://www.w3.org/2017/04/27-tt-irc
Attendees
Present
Glenn, Nigel, Mike, Andreas, Thierry, Pierre,
David_Ronca
Regrets
Chair
Nigel
Scribe
nigel
Contents
* [3]Topics
1. [4]This Meeting
2. [5]TTML
3. [6]TPAC 2017 planning.
4. [7]HDR in PNG
5. [8]IMSC
* [9]Summary of Action Items
* [10]Summary of Resolutions
__________________________________________________________
<scribe> scribe: nigel
This Meeting
Nigel: Today we have TPAC, IMSC, TTML, HDR in PNG.
David_Ronca: [requests TTML2 at the head of the meeting for
scheduling reasons]
Nigel: Anything else for the agenda, or AOB?
group: [no AOB]
TTML
Nigel: First let's look at progress tracking.
... First thank you Glenn for assigning yourself to issues. The
number you're assigned to is
... quite a bit larger than the number of branches you have
open. What's happening there?
Glenn: I've either begun work on them or thought about them and
have a plan. If someone
... wants to work on one and I don't have a branch for it then
coordinate with me and let
... me know. We can have more than one assignee for any issue.
Nigel: That's true we can.
David_Ronca: For those issues you have begun working on but do
not have a branch for it
... yet is there any reason why you wouldn't unassign yourself
so people can filter on
... assignee?
Glenn: If someone wants to take one on then just coordinate
with me.
Nigel: If you're assigned to issues then that will likely put
... other people off, and the flip side of that is that the
Group will expect some short term work on it.
Glenn: There are 5-10 issues with no assignee that I'm less
interested in, so people should
... start with that first.
David_Ronca: What I'm interested in most is not who resolves an
issue but when they are
... resolved. I see 9 issues unassigned and marked as milestone
TTML2WR, and 48 assigned
... to Glenn. Glenn, for the issues that are open and assigned,
is there any way to put an
... expected close date on them?
Glenn: I haven't looked at them with that level of granularity.
Nigel: I think the important thing is that we have pull
requests open for all those issues,
... as the first step towards issue closure.
... We talked about doing this by the end of May for all issues
for the WR - I've had feedback
... this week that before concluding on the WR we should review
all open issues so that we
... do not defer without due consideration, which is reasonable
I think.
Glenn: Some pull requests cover multiple issues.
Nigel: Please could you mention all the issues in the Pull
Request description so we can track those?
Glenn: I can add links but it may be difficult to uncover the
information.
Nigel: Has anyone else assigned themselves any issues?
Pierre: My plan is to assign myself all the horizontal review
comment issues and to resolve
... them before wide review. That can either be by pull request
or to defer until never or later.
Nigel: Okay, will you pick them up one at a time?
Pierre: Yes, I'll review them, hopefully with Richard, and
maybe close some of them in parallel
... or some sequentially. I don't plan to assign myself to them
all at once.
David_Ronca: Due to other commitments I haven't managed to meet
internally, but in the
... next week we'll all be back and will review and see what we
can assign internally to the team.
Pierre: I have a question re:
[11]Add an illustration of pinyin for ruby align
[11] https://github.com/w3c/ttml2/issues/274
Pierre: There, Glenn has a comment, so you have an opinion and
have thought about it, but
... you have not assigned it to yourself. Is it okay for me to
proceed?
Glenn: Yes, please do. I did not try to assign to myself any of
the i18n comments from the
... recent review, because they're not on the WR milestone.
Pierre: Why is that?
Glenn: Because they came in from outside the group so I
consider it part of the preliminary
... wide review comments.
... If I had no other issues to work on I would probably take
these on.
Pierre: There are other issues that other people can handle
that don't relate to complex
... script layout issues.
Glenn: Some of those others are in areas that I do have deep
expertise on.
David_Ronca: This is also something that Dae has worked on.
... How about we set a target for assigning issues?
Nigel: I'd like to treat this more like weekly sprints so
please assign yourself issues on a
... week by week basis so that we can track the curve as the
weeks go by and not block
... others from picking up issues because they're already
assigned, if that is ok.
David_Ronca: That's fine, I'll pick that up with the team next
week.
Nigel: Can anyone else let us know now about any issues they
can pick up and work on
... over the next week?
Pierre: What about the TTML1 issues?
Nigel: They're in scope too because they generally apply to
both TTML1 and TTML2.
Pierre: Here's a simple one:
[12]tts:extent="auto" vs tts:extent="100% 100%"
[12] https://github.com/w3c/ttml2/issues/215
Nigel: That's assigned to Glenn, and he said he is rewriting
that text anyway.
Pierre: That was in November.
Glenn: On this particular one I don't actually agree with the
proposed change. Auto is a
... proxy we often use to express a default without repeating
the definition of the default
... everywhere that proxy is used. I don't think there's any
technical reason not to use auto
... in place of 100% 100%. Also if we use auto then it's easier
to change what that means in
... one place, rather than everywhere it is used.
Nigel: If that's your thought process it would be helpful to
add it to the issue.
Glenn: Basically I consider this issue to be a request for
early binding of the extent.
Nigel: In this case maybe the preferred choice is to close the
issue without any change.
Glenn: The TTML1/TTML2 issue is an important one. In principle
we should have resolved
... all the open TTML1 issues and incorporate those resolutions
into TTML2 before PR,
... would anyone disagree with that?
Nigel: I think that's already agreed.
Pierre: Agreed.
Glenn: I just want to point out we don't have tracking issues
for every TTML2 issue that has
... a TTML1 issue, so there is more work than is indicated by
the currently open issues.
Nigel: How should we handle that, open tracking issues, or add
the TTML1 issues to the set
... we are tracking for TTML2?
Glenn: Or we could open an umbrella issue for TTML2.
Nigel: That could simply consist of a list of TTML1 issue
references.
Glenn: I think it would be useful to have that umbrella issue
at least to begin with.
Nigel: Shall I create that umbrella issue?
Glenn: Thanks.
David_Ronca: I have to drop off now - thanks a lot guys.
Pierre: How about:
[13]Step 10.4.4.2(6)(a) does not apply to textDecoration
[13] https://github.com/w3c/ttml2/issues/219
Pierre: This is discussed and agreed, really narrow, has had no
movement since London, so
... that would be a perfect candidate in my mind for anyone in
the group to take on.
Glenn: As I already mentioned, if you would like to add
yourself to an issue assigned to me,
... feel free to do that.
Nigel: can anyone take this on?
Glenn: We need to resolve this in TTML1 - coincidentally it is
#219 in TTML1.
... This comes under the PR for TTML1.
Pierre: I'd like to broaden the set of folk working on stuff.
... I'd get to it eventually but I'm going to prioritise the HR
comments.
Glenn: I'll get to it eventually as well.
<scribe> ACTION: nigel Create an umbrella issue for TTML1
issues in TTML2 [recorded in
[14]http://www.w3.org/2017/04/27-tt-minutes.html#action01]
[14] http://www.w3.org/2017/04/27-tt-minutes.html#action01
<trackbot> Created ACTION-496 - Create an umbrella issue for
ttml1 issues in ttml2 [on Nigel Megitt - due 2017-05-04].
Nigel: OK let's take that general approach - nobody is coming
forward for this now but I
... guess some folks' ears are burning but they want to think
about their workload and look
... at the open issues and see what they can take on.
... We'll come back to this again next week, and obviously
anyone can assign themself to an
... issue in the meantime.
... The next thing to do is to review the existing open pull
requests.
[15]Clarify application of padding (#221)
[15] https://github.com/w3c/ttml1/pull/233
Glenn: Andreas and I have gone back and forth on the utility of
adding implementation detail.
... I agree that we could say a little more because it doesn't
hurt to contextualise the subject
... with regards to the XSL-FO model. I'm not prepared to go
all the way to the level of detail
... that Andreas originally proposed.
... Looking at today's comment, Andreas said that how padding
is applied diverges from XSL-FO.
... I don't agree with that because, although XSL-FO does have
some semantics regarding
... regions and page layout semantics, the TTML definition of
region is entirely different.
... Maybe we need to point that out, in any case we are
defining something different and
... saying how it maps to XSL-FO, in §9.3.3 Synchronic Document
Construction.
... What it might be useful to do, as a subset of what Andreas
proposed, is to add a note that
... clarifies that the extent of region maps to the allocation
rectangle of the block area that
... is generated by the block container to which the TTML
region element is mapped.
... See also my comments on
[16]https://github.com/w3c/ttml1/pull/233#issuecomment-29763702
1
... In earlier comments I think Pierre had asked the question
of whether extent maps to
... padding rectangle or content rectangle, and I commented
that it is neither, but to the
... allocation rectangle from which those other rectangles are
derived. I'm prepared to add
... an additional comment with less detail than what Andreas
asked for but says what I just
... said but does attempt to clarify that mapping.
[16] https://github.com/w3c/ttml1/pull/233#issuecomment-297637021
Andreas: In response to the first point, I'm still of the
opinion that on this one area we
... really diverge from XSL-FO and that makes it hard to
explain how we relate to XSL-FO and
... also CSS. I was not referring to the region element in
XSL-FO, but to the way block areas
... are handled. The way TTML does it, defining the allocation
rectangle and then applying
... inset space has no correspondence in XSL-FO. You cannot
define the width and height
... of an allocation rectangle in XSL-FO, which makes it hard
to relate to and also tricky to
... implement using CSS models and concepts. We both expressed
our points, but the most
... important thing is that users get the right understanding.
I'm happy with any change
... that gives users the right understanding of how this works.
Nigel: Are you saying that Glenn's proposal would achieve that
or not?
Andreas: As I mentioned, I would like input from other members
of this Group too.
... If we can explain how it works clearly without referencing
the conceptual model then
... I am happy with that, but I want to hear other opinions
about it.
Glenn: I have an addendum to my opinion: another way to think
about it is to posit the
... existence of a container that the block container is a
child of, and then have the extent
... of the TTML region be assigned to the content rectangle of
that posited outer shell
... container. That can be done conceptually without actually
having to generate such an
... element in the flow object hierarchy. Or if one really
wants to do that in XSL-FO there's
... an element called staticContent that could potentially be
used for that purpose. A common
... use is for running headers and footers, except in XSL-FO
the content is repeated on every
... page, though the generated ISD outputs could be considered
to be pages in the XSL-FO
... sense, to make that leap of logic.
Andreas: I'm a bit worried that we make it even more
complicated using a different solution.
... There is a discussed and agreed label, so I assume that
members have read the original
... pull request and approved it. If we take on another edit
then there must at least be some
... feedback from the group expressing some kind of preference.
Pierre: TTML1 has text directly relating to XSL-FO. Couldn't
the clarification be put there?
... I understand that it might not be appropriate to put it far
away from where it is defined.
... Surely there should be a place to put it assuming it is not
false?
Glenn: I don't know if it is false or not - we could certainly
put the detail note in §9.3.3
... where the allocation rectangle is defined. I think the
current proposal is too detailed
... and application specific and I'm not prepared to accept it.
Different implementations can
... do different things.
Andreas: I'm willing to accept the edited pull request from
Glenn then. I read through it
... and also with the test cases provided by Pierre today it
might be enough to see how it
... works. The trouble with XSL-FO is people might not be
deeply into that, so its debatable
... how helpful it is. For me, if we use Glenn's edit and then
possibly add some more text to
... clarify that extent corresponds to the allocation rectangle
of the region then I'm fine with that.
Pierre: I'll file a separate issue because the same term
"region area" is used for overflow and
... I'm fairly certain that we want any clipping to occur on
the inset content area.
Andreas: There are different areas, but if we do not clarify
exactly which one we mean then
... it is not clear which applies, so some additional wording
would be helpful there.
Glenn: I assume that if clipping occurs then it is the content
rectangle.
Pierre: I think that's what everyone assumes, so let's file an
issue for that.
Glenn: I think I've heard a plan here to accept the pull
request plus an additional note on
... §9.3.3 right under the bit about TTML regions mapping to
block containers, in which I
... would say that it constrains the allocation rectangle. I
can do that. First I will update the
... pull request with a new commit on that branch adding that
note, and if the group accepts
... it then I'll go ahead and merge it. I'm asking that we
pre-accept that so we don't need
... another round of review.
Pierre: The process is you'll submit the PR and then there's a
3 day review.
Glenn: I'm asking for a quick merge.
Pierre: I need the 3 days.
Andreas: I do too.
Nigel: Thanks for the constructive conversation everyone.
Andreas: I have to drop off - I just want to check about the
agenda saying 60 minutes?
Nigel: The first line in the agenda should say 120 not 60 -
that's an oversight on my part, apologies.
Andreas: OK. [leaves]
Nigel: Let's take a 5 minute break... [5 minutes elapses] ...
and we're back
Glenn: I have one follow-up comment from before the break - I
just want to remind us that
... strictly editorial non-substantive changes do not require
the 3 day period according to
... the current process.
Pierre: I was not accusing you of being malicious by merging
early, but for the point of
... efficiency it is a lot more efficient to leave a review
period and deal with comments rather
... than open a new issue etc.
Glenn: My view is just the opposite about efficiency.
Nigel: Let's look at the next pull request:
[17]Issue 0237 css2 reference
[17] https://github.com/w3c/ttml1/pull/238
Glenn: I moved the reference to CSS2 to be in a Note and moved
the CSS2 reference to the
... Other References because it is clearly no longer possibly
normative.
... When it comes to merging this into TTML2 I need to
re-review in case there are other
... places where CSS2 has come in as a normative reference.
Nigel: Yes, it's worth doing that review.
... Looks good to me.
Glenn: This will merge into the ED of TTML1 and then I will
propose an equivalent pull request
... for TTML2.
Pierre: I will add a review approval too, to match the comment
I made.
Nigel: With 2x LGTM you could probably merge this now but it
would be fair in the process
... to wait until tomorrow Glenn.
[18]Issue 0203 add examples of ja features part 2
[18] https://github.com/w3c/ttml2/pull/273
Nigel: This has been open since March, what's going on...
... Okay we don't have Dae with us today, and the last status
is that Richard Ishida was
... looking at it. I will prompt him.
Glenn: This contains some examples of some of the new Ruby
functionality particularly
... Ruby overflow and Ruby Overhang and since I have not
reviewed it I do not know if it
... correct. I will first of all review it for correctness.
nigel: I've added a comment to the pull request.
[19]Revised tts:luminanceGain
[19] https://github.com/w3c/ttml2/pull/294
Nigel: I haven't had chance to review this yet but it would be
helpful to review and merge
... so I can merge the same fixes in to the other HDR pull
request.
Glenn: It looks like that incorporates Nigel's comments.
Nigel: I see that, yes great, looks good.
Glenn: Can I go ahead and merge?
Nigel: If you've reviewed then it would be good to add that to
the pull request.
Glenn: I've done that.
Nigel: I don't see why that can't be merged at this stage.
Thank you.
Glenn: I'll do that after the call.
Nigel: Once you've done that then I'll update my follow-on pull
request:
[20]Issue #298 sRGB on HLG
[20] https://github.com/w3c/ttml2/pull/299
Glenn: My only comment on that is that the newly added
reference doesn't really match
... the format that we've used with the other ISO references,
for example it includes the
... ISO reference within the title, whereas the others include
only the name portion. The
... other is that you've been extremely specific in the way you
have referenced the ISO number
... and the CIE number and the date, etc. and generally we have
not added dates or tried
... to be overly specific on these references. Mike raised an
issue on this recently that might
... bear on this point.
Nigel: That's:
[21]reference versioning
[21] https://github.com/w3c/ttml2/issues/293
Glenn: The first question is if we need both the ISO and the
CIE references?
Nigel: I think so since the same document exists in both
organisations.
Mike: Technically the CIE is a process document. They publish
jointly with ISO.
Glenn: I would suggest using the ISO reference, rather than the
CIE. Also you've used the
... key of CIE-XYZ. I like to keep those shorter.
Nigel: I'm happy to change the key to XYZ.
Glenn: If you do it that way you may be able to remove the CIE
portion of the reference.
... Also you refer to a specific year on the ISO document,
where we don't generally do that.
... For example we recently added CMAF without a year
reference.
Mike: The first question is does W3C have a policy for what an
unversioned reference means?
... If not then you need to put the version of the reference
in.
Thierry: I don't think W3C defines what an unversioned
reference means.
Glenn: Operationally speaking it has always been understood to
mean "the latest" but it is
... not explicit. When Mike says that if there is no version
then we must specify one. That
... has not been the practice in W3C or in TTML. For example we
don't specify years in
... ISO specification references. If we create a new policy to
reference versions then we are
... going to have to change a lot of references.
Mike: Certainly in TTML2, there's a lot of water under the
bridge in TTML1. I don't follow
... your argument - we have a version of CSS2 for example that
has been mentioned recently.
... My poster-child for this is a SMPTE document that
references UNIX tar and someone
... blindly updated the year on the ANSI standard and in fact
ANSI had entirely removed
... tar from that particular standard. You never know what you
will get in future versions.
Glenn: One question is if there is a stable link to a document.
For W3C specifications we
... generally use a dated link to the document on /TR, but
HTML5 points to the unversioned
... "latest" link in /TR. One of the issue is that some of the
linked URLs are not necessarily
... maintained by their organisations as stable URLs.
Mike: I think as a general rule no standards body (other than
W3C or IETF) maintains
... stable links. In fact it is quite difficult to get older
versions of things.
Glenn: Another thing I find interesting is that for IMSC1 we
discussed dot-dot vs dot
... versions. Thierry mentioned that it is assumed that newer
versions supersede older
... versions automatically. That surprised me. I've never
thought that. That might explain
... why W3C has sometimes not put dated versions in, but that
makes it a moving target
... as you have pointed out.
Mike: Yes, at some point you will get into trouble with
interoperability.
Glenn: For TTML2 we need to review all of the references and
decide if we want to augment
... them with version data.
Nigel: You could set a general rule of "the most recently
published one at the time of
... spec publication" but that would be hard to manage.
Mike: That's impossible.
Pierre: Yes that's too difficult.
... The way to go is to use dated versions except where we
specifically want to track changes,
... for example for registries.
Mike: That makes sense.
Glenn: Maybe the resolution for #293 is to go through all the
references and check them out.
... That's not one that I assigned myself to. Maybe someone
would like to do that.
Mike: If it is not the current version of the spec as it is
today then I'm not sure how we go
... about resolving that. For example with CSS2 vs CSS2.1 we
know there is a difference and
... there's a right answer. The thing that triggered it for me
is documents I know about - I
... can certainly fix those. I don't know about the W3C
references.
Glenn: I'm looking at the W3C references in TTML2 and it looks
as though all of them point
... to a dated version, so that shouldn't need changing. I have
always avoided the "latest" link.
Nigel: Can I assign this to you for those you can handle Mike?
Mike: Sure. For normative references especially we need to be
specific. It may be as simple
... as putting the date in. So I might be able to do this even
for the W3C references if the
... dates are already present.
Nigel: Ok I've changed the labels, added a comment and assigned
that issue.
... For the pull request #299 Glenn please could you add those
comments and I will
... process them?
Glenn: Okay.
[22]Issue 0296 inline block
[22] https://github.com/w3c/ttml2/pull/297
Nigel: We don't really have time to cover this today, I have
added some review comments
... and asked a question.
TPAC 2017 planning.
Nigel: Note I have completed the survey for the group, with
answers as described in the
... agenda - please let me know if you think any of those
answers need to be changed.
HDR in PNG
Pierre: How do we proceed to publishing that Note?
... I have forwarded it to Chris Lilley and have not heard back
so should we just put it up for
... official review on the reflector, asking for review, and
then we can go ahead?
Nigel: yes, that seems like a good way forward.
Glenn: And the audience for review is our working group?
Pierre: Correct.
Nigel: If anyone (in or out of this group) has an interest feel
free to point them to it.
Glenn: We don't have an obligation for review for the Note.
Pierre: I'd suggest starting a review clock for this.
Nigel: When would you close the review?
Pierre: Two weeks from today - if anyone asks for more time
then we can adjust.
... That would be May 12th.
Nigel: Okay, I'll send out a request to the group.
Action-495?
<trackbot> Action-495 -- Thierry Michel to Update the ttwg
homepage and publications pages for the new repos -- due
2017-04-20 -- OPEN
<trackbot>
[23]http://www.w3.org/AudioVideo/TT/tracker/actions/495
[23] http://www.w3.org/AudioVideo/TT/tracker/actions/495
Thierry: I'll update those pages to reference the HDR in PNG
document.
Nigel: I'd be very interested to know Chris Lilley's opinion
too.
Thierry: I will ping him again.
IMSC
Nigel: I think the point to raise here is about CR exit
criteria.
Pierre: Yes, I am also waiting for comments to come back. The
baseline is only to test
... the added features. I don't have a lineGap implementation.
It might be nice to have a
... vertical writing example too.
... I will ask Andreas about that.
... I don't expect a large test suite but it will be a
challenge to generate the examples.
Nigel: I know we updated our gstreamer implementation to do
lineGap but I do not know
... if it does vertical writing.
... We had a liaison response today, available at
[24]https://lists.w3.org/Archives/Member/member-tt/2017Apr/0004
.html
... The sender requested member-only visibility.
... I should also have mentioned under TTML2 that there was
some further TAG
... discussion, at:
[24] https://lists.w3.org/Archives/Member/member-tt/2017Apr/0004.html
[25]Timed Text Markup Language 2 (TTML2)
[25] https://github.com/w3ctag/design-reviews/issues/138
Pierre: I might be able to point the commenter at IMSC.js for a
practical example of the polyfill question.
Glenn: I would say that it is a subset, since practicality
might be arguable. Not "the practical subset" but "a practical
subset".
Nigel: I also want to acknowledge Pierre's message about the
test suite - thank you for that.
... Thank you everyone, we are slightly over time today,
apologies. [adjourns meeting]
s/Ok, the flip side of that is that is that if/If
Summary of Action Items
[NEW] ACTION: nigel Create an umbrella issue for TTML1 issues
in TTML2 [recorded in
[26]http://www.w3.org/2017/04/27-tt-minutes.html#action01]
[26] http://www.w3.org/2017/04/27-tt-minutes.html#action01
Summary of Resolutions
[End of minutes]
__________________________________________________________
Minutes formatted by David Booth's [27]scribe.perl version
1.152 ([28]CVS log)
$Date: 2017/04/27 16:28:54 $
[27] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
[28] http://dev.w3.org/cvsweb/2002/scribe/
----------------------------
http://www.bbc.co.uk
This e-mail (and any attachments) is confidential and may contain personal views which are not the views of the BBC unless specifically stated.
If you have received it in error, please delete it from your system.
Do not use, copy or disclose the information in any way nor act in reliance on it and notify the sender immediately.
Please note that the BBC monitors e-mails sent or received.
Further communication will signify your consent to this.
---------------------
Received on Thursday, 27 April 2017 16:29:55 UTC