See also: IRC log
<scribe> scribe: nigel
Nigel: I added to the agenda the
TPAC 2017 advanced planning, and we have TTML and
... IMSC topics. Any other things to cover?
Pierre: I'd like to discuss moving forward with the test vectors for IMSC 1.
Nigel: Great let's do that.
Nigel: I believe we already
agreed to meet at TPAC for at least 2 (and possibly 4)
days.
... I will need to complete a survey saying how many days we
want to meet for and if
... we wish to meet any other groups.
... TPAC this year is in Burlingame, 6-10 November.
https://www.w3.org/2017/11/TPAC/Overview.html
Nigel: I will also need to state
the days preferences, our flexibility, and any overlaps with
other groups to avoid.
... There is another big event in San Francisco at the same
time so flights and
... accommodation will be limited - advice is to book
early.
... Does anyone already know of any date preferences for our
meetings?
Andreas: I will need to check - if we meet for 4 days then I will probably not join TTWG on all those days.
Nigel: OK I will come back to
this next week - I have until mid-May to complete the
... questionnaire on behalf of the Group.
Nigel: Two things here: firstly
liaisons, secondly test suite.
... For practical reasons I was unable to send the final
liaison until yesterday but have
... now done so, with apologies to the last group to receive
the message, which gives them
... a bit less time to deal with it.
... The next thing is the test suite.
Pierre: There's been a source of
recurring complaint that there is no complete test suite
... for IMSC so what I've done over the past couple of weeks is
to combine tests from the TTML
... test suite and the IRT test suite, and licences on both
allow that. I have modified those
... files to be conformant IMSC 1 and to make them more
appropriate for generating test
... vectors. Then I ran IMSC.js on them and generated a
sequence of PNGs, one for each ISD
... based on them. I would like to submit those back to W3C and
make them the official
... IMSC test vectors. One obvious objective is to help folk
develop and test software, another
... is if results differ then we can review that in an issue,
in case it is an issue with the test,
... IMSC.js or the spec. In practice I would create a pull
request against the IMSC repo
... replacing the test subfolder with the test vectors.
Nigel: One question is whether we
base the IMSC 1.0.1 CR exit criteria on these tests or
... part of them or if it is completely separate?
Pierre: This submission is not intended for the exit criteria of IMSC 1.0.1 - it could be but that is not my primary goal.
Nigel: Does it include tests for the new features?
Pierre: Not yet, but I would hope to add them in time.
Mike: One of the new features has
no presentation output, so the test would be that it does
... not break.
Nigel: Quite right, though a
console output might log recognition of the activeArea.
... So the request is for W3C to host a test suite for
implementors beyond CR. Philippe, is that commonly done?
Philippe: Yes, we could do that,
but I would say that you should not put W3C staff in the
... critical path.
... I would recommend setting up a github repository for it and
serving it, then we can
... create a link to serve it from W3C with a redirect.
Nigel: So that repo could be the imsc repository, right?
Philippe: It could be but I would
not recommend it. The test suite licence would be
different
... so we do not need to track the IPR in the same way.
... I can create another repo in W3C - it is not complicated.
If you really want one repository,
... then be aware that any contributions from outside the group
will be tracked for IPR and
... you will not be able to accept them.
... We have licences for tests, so having two licences for one
repository would be a bit more
<plh> https://github.com/w3c/web-platform-tests/blob/master/LICENSE.md
<plh> https://github.com/w3c/web-platform-tests/blob/master/CONTRIBUTING.md
Philippe: complicated. For
example the web platform tests have a separate
repository.
... The examples above don't look at all like what is in the
IMSC repository.
... The more I think about it they should be in a separate repo
with a clearly different licence
... and contribution guideline from the spec.
Pierre: Then I would propose to remove the tests from the spec repository.
Nigel: Will it be okay to set CR exit criteria based on tests outside the spec repo?
Philippe: Of course, the tests
are not part of the spec.
... I recommend that we create a separate repo for this, and
then tell me where to link from
... in the W3C website and I will do it.
Pierre: I will take an action item to point you exactly to where they are.
Philippe: Thank you, I am happy to fork an existing repo.
Pierre: I will try to make it simple for you.
Philippe: It would be easier for
me to create the repo, and set the access rights the same
as
... the group's spec repos.
Pierre: OK.
<plh> https://github.com/w3c/imsc-tests
Philippe: That is a link to the repository.
Pierre: I will push the tests there.
Nigel: Great. Thank you!
... The existing CR implementation report is based on the tests
in the spec repo, so it
... wouldn't be great to remove them.
Pierre: Actually the IMSC 1
implementation report links to the Mercurial repo so nothing
we
... do here will affect the formal IMSC 1 implementation
report.
Nigel: Ok then I guess you're free to clean up the IMSC repo as you like!
Pierre: Thanks, I'll suggest something.
Nigel: Now we need to begin
thinking about the Exit Criteria for IMSC 1.0.1 CR.
... My starting point would be for two independent
implementations passing each of some
... new tests for the new features, which would be:
<plh> w3c/imsc-tests should be all set
Nigel: 1. Not failing when
processing a document with ittp:activeArea present
... 2. (perhaps) indicating on some console output the found
value of ittp:activeArea.
... 3. Successful presentation of the lineGap style as per the
example in the spec, in other
... words turning that example into a test vector.
... I think we should set the shortest period that we are able
to according to the Process,
... which I think is a month?
Philippe: For CR it is 4 weeks.
Nigel: Thanks.
... That would be my proposal. Does anyone think we need
anything more?
group: Nothing more needed.
Pierre: If you file an issue to remind me then I will add those.
https://github.com/w3c/imsc/issues/226
Nigel: Done, in above issue #226.
Nigel: Even in Glenn's absence I
would like to discuss our working process. He has sent me
... some written notes since he is unable to be present.
... Glenn's view as he's expressed it to me is that the ED is
the Editor's prerogative to
... work with, and that it does not necessarily represent the
Group's consensus.
... I would rather not work against each other here but work in
harmony. However we do
<plh> https://www.w3.org/AudioVideo/TT/IMSC/tests/
Nigel: seem to want a group
consensus so one option is to create a new branch to
serve
... as the group's consensus and publish WDs based on that.
Pierre: This is also an issue
because the ED is linked from the version on /TR.
... Another issue is that if someone creates a PR based on an
issue, especially if they have
... been asked to, then the discussion on the PR should happen
on that PR, so that everyone
... can track what is changing. It is unreasonable to create a
separate pull request and have
... the discussion somewhere else, or no discussion in the case
of an early merge, because
... it makes it hard to compare with the original
submission.
Andreas: Also, to reflect the
earlier discussions, this relates to the position of the Editor
in
... the TTML1 and TTML2 activity, where the Editor has an
extremely strong role. In my
... experience this is even a bit stronger than usual, and
maybe is not helpful for the group
... dynamic, so we need to agree how the different roles work
some time.
Nigel: In my role as Chair, I
think I need to identify when there is or is not consensus.
Right
... now it is clear that there are changes being merged into
the ED for which there is not
... consensus in the Group.
... Glenn's view on how to proceed is to raise issues on the
merged spec. I think we
... should recognise that from a perspective of the end result
document that could result
... in an okay end result, even if it is not a happy
process.
Mike: Looking at this from afar, it seems like we are not all agreed on the process we follow.
Nigel: Exactly, I would like to
come up with some kind of action plan to resolve this.
... Philippe, have you seen any other examples from other
groups that we could take
... inspiration from?
Philippe: If this doesn't work
for you guys in the Group then it needs to change, that's
the
... bottom line. If the group wants to review stuff done in the
Editor's own branch then that's
... okay.
Nigel: Before I make any proposal are there any other thoughts for how to resolve this?
Mike: Step one is to talk to the
Editor, and if that does not work then step two is for
the
... Chair to take control of merges to the draft.
Philippe: Yes.
... There is an easy way - we can change permission settings to
restrict who can merge
... into the repository.
Nigel: Okay, that's fine. And
regarding the point about the ownership of the ED, is
there
... anything in W3C or the Process about that?
Philippe: To be clear in lots of
other groups the ED does not represent the consensus of
... the group, even if the ED is linked from the /TR
page.
... Formal consensus is only represented in the WD.
Pierre: So do the CSS editors try to capture group consensus in the ED?
Philippe: No they don't.
Pierre: And are the changes to the ED made with pull requests or done unilaterally by the Editor.
Philippe: I would have to check.
It used to be that they would be done unilaterally by the
Editor.
... In some cases the Group would ask for changes on moving to
WD.
Pierre: The great thing about pull requests is you don't have to do that.
Philippe: I agree, but if you do not have that working method then it does not apply.
Mike: The goal is for the ED to
reflect the current consensus. Forcing the members to try
to
... correct something that was not originally proposed is a lot
of work that perhaps is not
... necessary. It is odd at best to have the editor to do
something different to what is in WD.
... I believe it would be a better process to use pull requests
and merge on consensus
... into the ED.
Philippe: Nigel, you're right,
the Process does not make any requirements on the ED.
... If the editor cannot produce a draft that the group is
willing to agree on then you may
... need a different Editor.
Nigel: Okay we've discussed this
enough today. In Glenn's absence I will not impose the
... option to stop early merge as indicated in the agenda,
however I will take an action to
... talk to Glenn about this situation. Something will change
one way or the other.
... In Glenn's absence are there any TTML issues that anyone
wants to raise?
Andreas: I have to drop off the call now.
Nigel: The only one from my list on the agenda that I think we can easily cover is:
There's no apparent use case for the "any" term in ttp:contentProfiles
Nigel: I agree with Pierre (who
raised the issue) that there's no meaning for "any" here
and
... we should simply remove it.
Pierre: That was like 2 years ago.
Nigel: I've added a comment on
the issue.
... I've also labelled it as "Discussed and agreed".
Pierre: I'd like to look at 244
-> https://github.com/w3c/ttml2/issues/244 Signaling HDR pixels in PNG for use with <image>
Pierre: I'm sympathetic to the
point that TTML2 does not define any image formats.
... However the proposal raised here is something the industry
will do regardless, i.e. to put
... PQ pixels in PNGs. Noting this in TTML2 seems reasonable
because it is most likely to
... happen in subtitles, and because W3C is the home of PNG. I
had suggested adding it
... as an informative annex in TTML2. An alternative perhaps is
to create a WG Note, but I
... think it is important to write it down somewhere otherwise
people will do crazy stuff
... all differently. The advantage of a WG Note is that if
someone outside this Group wants
... to make it more formal then it is easier to pull out. The
drawback is it is further away
... from TTML2 though there could be an informative pointer to
the WG Note.
... I think it is important to point to something but I'm not
all set on it being in the TTML2
... spec.
Nigel: From my perspective, it is
easier to get consensus on a WG note and it does not need
... to hold up TTML2, so I think if you want to draft something
then go ahead.
... Is a problem here that it is a misuse of PNG in some
way?
Pierre: That's a really big
question, on review I think not. PNG allows for arbitrary
... gamma exponents but it only supports power law EOTFs and
also arbitrary ICC color profiles,
... so it is not clearly an abuse of PNG, just not using one of
the built in EOTF gamma curves
... defined today, instead using an ICC which is permitted. I
think this is classed as a
... permitted extension. It may be better if PNG were to
support EOTFs other than just gamma,
... but that's something that others in W3C may want to work
on.
Mike: Would changes to PNG be made in ISO or W3C?
Philippe: It would be W3C. It
certainly has been developed jointly in the past with ISO. I
can
... even give you the right contact in W3C.
Pierre: Maybe the sequence of
events is to draft the WG Note and then see if there is
further
... interest.
Philippe: Chris Lilley created the Color CG with Mark Watson - Chris is the guy to talk to
<plh> https://github.com/w3c/strategy/issues/58
Philippe: in any case and he is also the one tracking our PNG stuff such as the above issue.
Nigel: WG Notes can be turned into Recs later, right?
Philippe: Yes they can or they can be parked.
Pierre: My position is we should
publish the WG Note and then make it available to the Color
CG,
... and even bring Chris Lilley into that work. That makes it
really concrete.
Nigel: I'm happy with that.
Mike: I don't have a particular
plan in mind I just wanted to raise the question and see
if
... there is another group we should collaborate on it. This
plan to publish a WG Note and
... then work with others is a good plan.
Pierre: Ideally PNG would be
updated to take into account these new use cases, but
that
... would take some work to get sufficient critical mass.
Nigel: I've added a comment to the issue on this.
Pierre: Could we create a repo for this?
Philippe: Sure, what name do you want?
Group: [some discussion] png-hdr-pq
Nigel: That allows for other HDR
extensions to PNG, if needed, to be placed in other WG
Notes
... whose repositories would be called e.g. png-hdr-??? where
??? would be replaced.
Philippe: I'll get onto that.
Nigel: Thank you!
<plh> https://github.com/w3c/png-hdr-pq
Nigel: I've added a link to that
and closed the issue - any request to create a reference
to
... the WG Note should be done in a separate new issue.
Pierre: I'll use Respec for this.
Philippe: I'll set it up for the
WG Note.
... I will also set up the master branch to be the one that
github.io serves.
Pierre: That's great.
Philippe: I did the same thing
for the imsc-tests repo as well.
... And it is replicated on the W3C website.
<scribe> ACTION: tmichel Update the TTWG homepage and publications pages for the new repos [recorded in http://www.w3.org/2017/04/13-tt-minutes.html#action01]
<trackbot> Created ACTION-495 - Update the ttwg homepage and publications pages for the new repos [on Thierry Michel - due 2017-04-20].
Nigel: I think we've run out of steam for today. Thanks all. We have 2 hours again next week. [adjourns meeting]