- From: Nigel Megitt <nigel.megitt@bbc.co.uk>
- Date: Thu, 2 Nov 2017 15:52:17 +0000
- To: Timed Text Working Group <public-tt@w3.org>
- Message-ID: <D620EF58.4D412%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/2017/11/02-tt-minutes.html
In text format:
[1]W3C
[1] http://www.w3.org/
Timed Text Working Group Teleconference
02 Nov 2017
Attendees
Present
Nigel, Pierre, Glenn, Cyril
Regrets
Andreas, Thierry
Chair
Nigel
Scribe
nigel
Contents
* [2]Topics
1. [3]This Meeting
2. [4]TPAC 2017 Advanced planning
3. [5]TTML1 and TTML2 compatibility
4. [6]TTML2 Pull Requests
5. [7]CSS mapping for tts:fontShear
6. [8]document proceessing context override semantics
pull request #468
7. [9]Are nested ruby annotations needed? #476
8. [10]contentProfiles and processorProfiles delimiters
possibly ambiguous #474
9. [11]ttp:profile has been extended from TTML1, but no
new feature has been defined #473
10. [12]TTML style mapping into CSS features
11. [13]Processor backwards compatibility. #458
12. [14]Is #writingMode-horizontal too general in image
profile? #147
13. [15]IMSC 1.1 profile signaling and resolution
14. [16]Example of documents that fail the HRM #71
15. [17]Add HRM threshold test(s) #37
16. [18]Meeting Close
* [19]Summary of Action Items
* [20]Summary of Resolutions
__________________________________________________________
<scribe> scribe: nigel
This Meeting
Glenn: [needs to go after 1.5 hours]
Nigel: [would also like to finish promptly] I've been invited
to attend the PING meeting
... immediately following this one to discuss the call for
horizontal review of IMSC. Others
... are welcome too.
... Today I'd like to review the schedule for next week's F2F
meeting.
... In the last week we've had some development on IMSC, and
some discussed issues
... for TTML but not much change. There's been some issue
discussion on WebVTT too.
... I've received no hints from David, Silvia et al that they
will join to discuss WebVTT issues.
... Any specific points to raise today?
Pierre: 2 issues on IMSC 1.0.1 that we're waiting on folks from
i18n on so we need to find
... a way to reach out to them somehow.
Nigel: OK - maybe the easiest way will be to try to grab them
in person next week.
Pierre: Okay - last time Thierry contacting them directly
worked well too.
Nigel: If it can wait until next week that would be best I
think.
group: [no other points to raise specifically today]
TPAC 2017 Advanced planning
[21]TPAC TTWG wiki page
[21] https://www.w3.org/wiki/TimedText/tpac2017
Nigel: When is good for Australia?
Pierre: Their day usually starts around 3pm California time.
Nigel: I'm guessing David Singer will be at the AC on Thursday
so I'll schedule WebVTT for the Friday afternoon.
... [edits schedule]
Pierre: We ought to try to cover how to integrating IMSC 1.0.1
features in TTML2, e.g. fillLineGap.
... We haven't managed to solve it offline.
Nigel: At the moment the schedule looks like:
... Thursday am: IMSC
... Thursday pm: TTML2
... Friday am: slack time plus CSS WG joint meeting
... Friday pm: Charter, TTML <--> WebVTT mapping, WebVTT WR
comments
Cyril: I will have a call to make on the Thursday morning. I
will check if Netflix can move things around.
... I will let you know more later today.
TTML1 and TTML2 compatibility
Pierre: What's the plan for TTML1 Third Edition and who is
working on it?
Nigel: Last week Glenn told us that he plans to do TTML1 TE but
hasn't made any commits
... since May, and would welcome other editorial input.
Glenn: There are few technical issues if any so it's just a
matter of putting the pull requests
... in.
Pierre: What's the schedule for it?
Glenn: It's lower priority on my worklist than TTML2 at
present.
... I wouldn't mind having other editors take that on.
Pierre: I'll investigate if I can contribute.
... I see that all the Third Ed issues are labelled as such.
Nigel: I think they're in a Milestone.
... One other useful thing is that we may be able to diff TTML1
SE against TTML2 to pull out
... fixes for specific issues.
Pierre: That's a good idea.
Nigel: I don't think there are any updates on the compatibility
document?
Cyril: The only updates are I added links to issues.
... I will try to move the document onto the wiki.
Nigel: Thank you that would be really helpful.
Glenn: On the TTML1 repo, there are some issues open on the
test suite, labelled as Third Edition Test suite milestone - 11
open.
... And there are 42 open issues against the spec itself. Some
of them have a cross-dependency
... on TTML2 so I will probably be able to work on those to
make sure they get done on
... both sides. First order, having someone looking at the 11
test suite items would be very useful.
Pierre: This is lower priority I think.
Glenn: I agree.
Nigel: I think doing the spec first and then addressing the
tests makes sense.
Glenn: There are 14 Discussed and Agreed so hopefully they can
be resolved fairly quickly.
... They would be the first low hanging fruit to work with.
Nigel: There's only one bug that is on the spec and is not
discussed and agreed.
[22]Inconsistent implicit duration of singleton span in
sequential container. #193
[22] https://github.com/w3c/ttml1/issues/193
Nigel: It looks like in January you asked to check it again
Glenn.
Glenn: Yes, I think I reported it originally.
... And it's marked as applying to TTML2 as well.
Nigel: By the way, as Chair, just a reminder that any group
member can submit pull requests
... - you don't have to be an Editor.
... Plus as Chair, and in general in W3C, it's good to have
more than one Editor per spec,
... since it derisks succession planning and resource
bottlenecks etc. so I'm always keen
... to receive offers to volunteer as an Editor.
TTML2 Pull Requests
[23]Issue 0406 style semantics #470
[23] https://github.com/w3c/ttml2/pull/470
Glenn: Last week we discussed this - I need to finish reviewing
this, and will try to get it
... done in the next day or so.
Nigel: The only thing I'm aware that's substantive that is
missing is fontShear.
... Cyril, you said you would check my interpretation of
tts:fontShear mapping to CSS transform skewX() and skewY()
functions is correct.
Glenn: FYI that's how I implemented it in TTPE, using SVG's
skew transformations. If CSS
... supports that then it would be a good approach.
CSS mapping for tts:fontShear
github: [24]https://github.com/w3c/ttml2/issues/471
[24] https://github.com/w3c/ttml2/issues/471
Nigel: Cyril, you said you would check my interpretation of
tts:fontShear mapping to CSS transform skewX() and skewY()
functions is correct.
Glenn: FYI that's how I implemented it in TTPE, using SVG's
skew transformations. If CSS
... I also used rotation because certain characters in vertical
mode are labelled as
... requiring rotation by the Unicode technical report on
vertical layout.
Nigel: Is that part of fontShear?
Glenn: The skew functions are for fontShear, the second is for
vertical layout.
Nigel: Cyril, Glenn, if you could add a note on the issue
confirming I've mapped the inline
... progression direction to the correct skew function that
would be helpful.
document proceessing context override semantics pull request #468
Glenn: I need to update the pull request for two reasons
(editorially) so I believe I can do that
... and then implement the change.
Nigel: I think that's fine, we discussed this last week.
Are nested ruby annotations needed? #476
github: [25]https://github.com/w3c/ttml2/issues/476
[25] https://github.com/w3c/ttml2/issues/476
Glenn: Pierre asked a good question, if nested ruby is going to
use. I don't know of anybody
... implementing or requesting this feature. It's in the spec
because it is a technical possibility
... based on the way the Ruby markup works. I believe this is
also true in HTML ruby. I haven't
... seen restrictions on this in HTML or CSS ruby specs.
... It is technically feasible, and I could imagine creating
content that would use it but its
... edge case and I could not imagine using it in a subtitling
or timed text mode at the moment.
... My expectation is it will fall out of the CR process, so we
should at least mark it as at risk in the CR.
Pierre: So just to confirm, Glenn, you're not aware of any
Timed Text application for it?
Glenn: I've seen it in typesetting in print but not in
subtitling. So it's theoretically feasible.
Pierre: How does it look in print?
Glenn: For example, in Japanese you can have kanji that
annotates other kanji with ruby on
... the second level of kanji. For example a semantic
annotation in kanji and then a phonetic
... annotation of that semantic annotation in hiragana or
katakana.
... It just happens that TTPE supports it but I haven't created
any tests for it.
Pierre: Thanks for clarifying.
Glenn: It could be in the JLReq.
Pierre: I searched for it.
Glenn: In the mid-1990s it was a required feature in the work I
did on Japanese typography.
... It was an edge case though.
Pierre: Alright, then the question is if it is used for
subtitles and captions.
Glenn: My proposal right now is to plan to mark it as at risk.
Nigel: I noticed on a recent other CR that the exit criteria
were 2 implementations for required feature
<cyril>
[26]https://html.spec.whatwg.org/multipage/text-level-semantics
.html#the-ruby-element
[26] https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-ruby-element
Nigel: and 1 implementation for optional features, so that's an
interesting option for us.
Cyril: The above link - if you scroll down to "Text with both
phonetic and semantic annotations (double-sided ruby) "
... then the section includes nested rubys.
... §4.5.10
<pal> [27]http://fantasai.inkedblade.net/weblog/2011/ruby/
[27] http://fantasai.inkedblade.net/weblog/2011/ruby/
Pierre: I found another example that is syntactically nested
but one ruby is above the base
... and the other is below.
... How would we achieve annotations both above and below the
base in TTML2?
[28]Ruby document by fantasai
[28] http://fantasai.inkedblade.net/weblog/2011/ruby/
Nigel: Scroll down to the "Double Ruby" section.
Glenn: That's not ruby on ruby.
Pierre: The markup is nested but the result is not.
Glenn: Right now it would follow that "San Francisco" double
ruby example. There are 3
... base characters that could have been wrapped in rb, and
there's a ruby around the entire text.
... I haven't tested that in TTPE. Syntactically, it has a
container within a container.
Pierre: How would you do that in TTML2 today?
Glenn: You'd have a span tts:ruby="container" with two
children, one another span tts:ruby="container"
... and the other with tts:ruby="text"
Nigel: So it is the same but with ruby mode expressed in spans.
<cyril>
[29]https://developer.mozilla.org/en-US/docs/Web/HTML/Element/r
tc
[29] https://developer.mozilla.org/en-US/docs/Web/HTML/Element/rtc
Cyril: The Mozilla documentation above does exactly the same
thing using rtc.
Pierre: So the Mozilla example is not nested syntactically.
Glenn: Right now having nested container spans is prevented by
one of the constraints in TTML2.
... So I wonder if nested ruby is actually supported. I'm going
to need to check this.
SUMMARY: @skynavga to check current nested ruby status in TTML2
contentProfiles and processorProfiles delimiters possibly ambiguous
#474
github: [30]https://github.com/w3c/ttml2/issues/474
[30] https://github.com/w3c/ttml2/issues/474
Glenn: I can mark that as discussed and agreed?
Nigel: Yes we did agree on the approach.
Glenn: [marks as discussed and agreed] and [editorial]
ttp:profile has been extended from TTML1, but no new feature has been
defined #473
github: [31]https://github.com/w3c/ttml2/issues/473
[31] https://github.com/w3c/ttml2/issues/473
Glenn: Yes, I agree with Pierre that we need to do this.
Nigel: Pierre, you asked for a single feature, right?
Glenn: This is on the ttp:profile element not the attribute.
Nigel: Should it be one feature or three features?
Glenn: The simplest thing would be one feature like
#profile-version-2 or whatever convention I use.
... Right now #profile is an aggregate of the features that
refer to five different elements.
... I think probably the best thing would be to collect all of
the upgrades to those elements
... and put them in a #profile-version-2 feature but I'd be
open if someone feels we need
... more fine-grained features.
Nigel: I'd consider testability here.
Glenn: If you're implementing the new TTML2 semantics in this
area you're really going to want
... to do everything that's been added rather than picking and
choosing, so my suggestion
... would be to tentatively aggregate them into a single new
feature designator.
Nigel: That seems reasonable
group: [no counter statements]
SUMMARY: Group tentatively agrees to create a single new
feature designator for this.
TTML style mapping into CSS features
Glenn: Nigel did you make progress on tts:wrapOption?
... I was surprised it was complex because wrapOption came from
CSS.
Nigel: I can't remember where I'm up to with that but I think
the processing of white space
... came into this as adding complexity.
... The CSS white-space property controls white space handling
so that interacts with xml:space.
Glenn: We don't support all the variability in XSL or CSS in
this regard and just map xml:space
... to a particular set of properties that then flow through.
... So to the extent that wrapOption semantics are dependent on
xml:space then ...
[misses details]
Nigel: If we want xml:space="preserve" and
tts:wrapOption="noWrap" then they are two
... mutually exclusive values of CSS white-space.
Glenn: Okay... [investigates]
[32]CSS white-space property definition
[32] https://drafts.csswg.org/css-text-3/#propdef-white-space
Glenn: I imagine that one or more of those XSL properties not
in CSS somehow map to the
... CSS flavour of the white-space property.
Nigel: I hope so!
Glenn: Let's take this offline.
Nigel: Agree!
... Other inputs very much welcome.
Processor backwards compatibility. #458
github: [33]https://github.com/w3c/ttml2/issues/458
[33] https://github.com/w3c/ttml2/issues/458
Cyril: I proposed to replace the entire §3.4.2 section.
Nigel: That looks good to me. Did you intend to include the
editor's note?
Cyril: We don't need to if we have issues recorded somewhere.
... I know we agreed to have something in TTML1 Third Edition,
so then if we record the
... issues (which I'll make sure they are) then we can remove
the editor's note.
... Shall I submit a pull request?
Glenn: I don't think we need to have the phrase "TTML2 is
designed such that". Why don't
... we just change "is intended to be" to "is"... Can I propose
a PR for these that's maybe a
... little less wordy?
Cyril: Sure!
Glenn: Okay I'll do that today.
SUMMARY: @skynavga to propose a pull request based on
@cconcolato's proposal
Is #writingMode-horizontal too general in image profile? #147
github: [34]https://github.com/w3c/imsc/issues/147
[34] https://github.com/w3c/imsc/issues/147
Pierre: this is super-boring and has a pull request open.
Nigel: I did ask myself the question of if we decide we need a
mixed text and image profile
... then if we would need horizontal-rl, and thought we
probably would.
Pierre: Absolutely, this is just on the image only profile.
Nigel: I concluded we don't need it on image only also.
IMSC 1.1 profile signaling and resolution
Glenn: Is it possible to infer an EBU-TT-D processor profile?
Pierre: yes using the fallbacks in TTML2.
Glenn: OK
Pierre: If you start with only ttp:contentProfile you always
end up with a processor profile.
... That's why EBU-TT-D ends up as the processor profile in the
TTML2 algorithm.
Glenn: In that case where the document interchange context
provides the profile, not that
... EBU-TT-D actually publishes a processor profile document.
Pierre: That is correct.
Glenn: It would be nice for them to do it [publish a processor
profile document].
Pierre: I don't think it's necessary for this to work.
Glenn: .. for interoperability reasons.
Nigel: I will take a look at this pull request (w3c/imsc#267)!
Example of documents that fail the HRM #71
github: [35]https://github.com/w3c/imsc/issues/71
[35] https://github.com/w3c/imsc/issues/71
Pierre: I've added an example that fails. But as a stylistic
issue I don't like to put failing
... examples in specifications. I'd rather add something to the
IMSC test suite that provides
... documents at the threshold of the HRM. My inclination is to
close this issue and add to
... the imsc-tests suite documents ones that are at the
threshold. I think that achieves the
... same objective, first providing a test to check that
implementations meet the HRM requirements,
... and secondly allowing people to check they have the same
understanding of the HRM.
Glenn: I have a different opinion philosophically - counter
examples are an important part
... of any formal specification. So I think it's very useful to
have official counter-examples.
Pierre: Maybe in the test suite.
Nigel: In the test suite it is a positive test for a validating
processor that it should reject
... documents that do not pass the HRM test.
Pierre: I'm concerned about spec readers assuming that the
document is a good one even
... though it is a bad one.
... In the test suite we could have both documents that fail
and documents that are on the threshold.
Glenn: And you might put the failures in a separate folder.
Pierre: Definitely!
... My proposal is to close this issue.
Nigel: Is it okay by you Glenn to have this in tests only?
Glenn: Yes
Cyril: Agreed
RESOLUTION: Close this issue, in favour of adding threshold and
fail tests to imsc-tests.
Add HRM threshold test(s) #37
github: [36]https://github.com/w3c/imsc-tests/issues/37
[36] https://github.com/w3c/imsc-tests/issues/37
SUMMARY: As discussed in the IMSC spec w3c/imsc#71, we will add
HRM threshold and fail tests which are positive tests for
validating processors.
Meeting Close
Nigel: Thanks all, we'll close now - see you all next week.
[adjourns meeting]
Summary of Action Items
Summary of Resolutions
1. [37]Close this issue, in favour of adding threshold and
fail tests to imsc-tests.
[End of minutes]
__________________________________________________________
Minutes formatted by David Booth's [38]scribe.perl version
1.152 ([39]CVS log)
$Date: 2017/11/02 15:49:44 $
[38] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
[39] http://dev.w3.org/cvsweb/2002/scribe/
Received on Thursday, 2 November 2017 15:52:53 UTC