See also: IRC log
<scribe> scribe: nigel
Cyril: I can stay only for an hour today
Nigel: Today we have CfC Status
update, Implementation Report run-through on each of
... TTML1 3rd Edition, TTML2 and IMSC 1.1, and any other open
agenda issues.
... There's one I know of which is for audio features on
TTML2.
... Then we can also cover IMSC vNext Requirements and I'm not
aware of anything else.
... Any other points to cover, or other business?
Glenn: When we get to TTML2 I'd like to get approval for a pull request that's outstanding.
Nigel: Okay let's look at that in
the TTML2 agenda item.
... By the way I haven't put TPAC agenda on the meeting agenda
yet, we need to begin
... thinking more about that in the coming weeks.
... By the way there is a wiki page for TPAC so please add your
names if you have not already,
... and if you intend to attend.
Nigel: As per the plan we agreed
I issued CfCs for TTML1 3rd Edition, TTML2 and IMSC 1.1
... yesterday, which began the review period.
... Please do check the documents are suitable for publication
as PR. We won't request the
... transition to PR until we have completed the Implementation
Reports.
Glenn: Please could you remind us of what date we have to submit the transition request by?
Nigel: Yes, we are on the "TTML2"
column for timing for all three specs now.
... That means end of CfC 26th September, IR complete 27th
September, Staff checks
... and uploads specs to final destination on TR on 28th
September, and PR transition request
... submitted on 28th September.
... One thing to note about the CfCs is the SoTD sections have
been updated for PR,
... and need to be checked.
... Thierry, please could you check the AC review link is the
right one?
Thierry: Yes, I'll check that.
Nigel: Thanks. Another thing is
for TTML2 there may be a feature removal/adjustment to
... be made, and there seems to be one other pull request to
review.
... On the whole though, we're in a pretty good state, the
state we said we'd be in!
... Thank you everyone for your hard work getting us to
here.
... A few more leaps to make of course!
... Any comments on those CfCs?
group: [silence]
TTML1 3rd Edition Implementation Report
Nigel: I see it is partially
populated at the moment, with entries for imscJS and
empty
... columns for TTPE and ttval.
Glenn: Please could we review the CR exit criteria?
Nigel: Two independent implementations
Glenn: So if I add TTV or TTX in would that be adequate?
Nigel: Yes
Pierre: One column or two separate ones?
Glenn: Separate ones. We can
leave the TTPE one in place and mark the two value font
... size tests as passed. I may have a chance to verify the
other ones and will attempt to
... do so on the TTPE column.
... I will run all the tests through TTX and TTV to make sure
everything is kosher there, shortly.
Pierre: Just to confirm, in the
case of the anamorphic font tests, there will be one
... presentation engine implementation and there will be one or
more validator implementations.
Glenn: Pierre, will you be filling in the ttval column?
Pierre: I will do for at least the two value font size tests.
Nigel: I'm a little uncertain
still after getting a response to my advance notice to the
Director
... about our approach to demonstrating implementations of
those features. The reason
... for my uncertainty at this stage is because the spec
changes did not affect validation
... but only affected computation of the font size.
... So a mere validation test could arguably offer no
demonstration of implementability of
... the change since TTML1 2ed.
... The Director seems to want something that does show that.
It could be that a validator
... is okay, but certainly not a simple XSD schema runner.
Pierre: The exit criteria is
clear - just 2 implementations, without qualifying what kind
of
... implementations. I'm trying to understand what were the
exact concerns of the Director.
... Moving the goalposts at this point is unreasonable.
Glenn: Before we dive into this,
I don't think we're depending on what the Director is
... worried about. We have a pass on presentation in TTPE for
those two tests and I think
... the ttval column will be ticked. Neither ttval or TTPE make
use of schema only validators
... anyway so that concern is not applicable.
group: [discussion of the minimum
requirements for demonstrating that each test has
... passed]
Pierre: Note that changes to
TTML1 text are not in the test suite for TTML2, because
we
... will not test them twice.
Glenn: Yes, the TTML1 text changes are matched in TTML2, deliberately.
Nigel: As set out above we
effectively have 14 days from today to complete the
implementation
... report. Is there any reason we would not be able to do
that?
group: [no reason to say no]
Nigel: There are no issues or
pull requests marked for the agenda.
... Just want to call out that pull request #367 is for the PR
version. I knocked it together
... yesterday to be able to get the CfC out while America was
asleep. So please do review
... the changes made there and the SoTD in particular.
Pierre: Looks good, I'm just merging in the CR exit criteria tests. Thanks for doing that.
Nigel: No problem.
<glenn> https://docs.google.com/spreadsheets/d/1gtUAV0r4Zd1NUSEcFclBqOrk41pOyaOTfaLxc7boBeQ/edit#gid=0
TTML2 Implementation Report wiki page
Nigel: I see Glenn just posted
the Google spreadsheet that we're using as a work in
progress
... Thanks for that.
... The #T column colour code shows if we are passing or
failing on CR exit criteria.
... There are a few of those.
Glenn: The one that have yellow
in the TTPE column, I'll be taking the yellow out and
those
... will all be turning into normal Xs so that will cause those
items to be green on the #T
... column, which is my final flag for passing the exit
criteria at this point.
... I've started updating the total numbers and the numbers on
the right side because there
... have been some changes in the IRT Subcheck column and the
NFLX-V column where
... previously an X was reported and now it is an S which is
described in the readme file
... of the repository.
... S means "strictly passes", i.e. does not produce any false
negative on validity tests.
... F means "fully passes", i.e. S but also does not fail to
report any invalidity on the invalidity tests,
... so there are no false positive tests on the
invalidity.
... I am counting only those marked F and discounting S as
equivalent to "partial" at this
... point. I don't think there will be an issue from doing
this. I need to update the totals
... to make sure they reflect the change to S on some of those
entries.
... The reason I'm fairly confident that there's not going to
be an issue there is that TTV
... reports F on all entries and ttval reports F on most except
for a few like audio features,
... luminanceGain and disparity.
... That basically leaves unresolved in my mind the issue you
posted Nigel on the audio
... features and whether or not we are going to be able to
check all those off.
... There were a few that you feel may need to come out because
you can't report a
... positive implementation on those, right?
Nigel: Yes, here's an update on
the audio features.
... First, I've just opened a pull request, just before this
meeting, adding a set of audio tests
... including all the audio styling attributes and applying
them to p, span, audio and animate
... elements, so that should assist with demonstrating those
features, especially if they
... pass validation.
... If anyone can test them for validity that would be
helpful.
Glenn: I will run those through the validator. I might need to add a Wave format validator.
Nigel: We don't need to validate the WAV file format.
Glenn: Unfortunately the way TTV
works I may need to add a simple validator to check
... that a WAV file is ok.
Nigel: Thank you.
... The next part of the update is regarding implementation
itself.
... I have good news in the sense that I've managed to steer
the oil tanker of the BBC to
... apply some development effort to completing this, which
should be complete by Sep 21
... which is within the timescales we have to work in.
Cyril: Quick update on TTML1 3rd
Ed - we have an implementation that passes the two
... value font size presentation test.
group: [general happiness]
Nigel: Back to the audio feature
implementation.
... I expect our implementation to pass on pan, gain, speak and
pitch.
... I haven't created an embedded audio test, but if time
allows we will try to implement it.
... We will try to implement #embedded-audio.
Pierre: Should we prepare a pull
request that removes those features in case we don't
... pass the CR exit criteria for those features?
Glenn: I can prepare a pull request but would rather wait until 22nd to do that.
Nigel: Yes, we should have finished by the 21st.
Pierre: Works for me.
Nigel: We need to get any pull
requests merged during the CfC period, so we should aim
... to merge such a pull request before the end of the CfC
period.
... Also I want to point out that in general we want specs to
be stable during CfC but in
... this case, anticipating the potential change, I flagged it
as a possibility in the CfC.
... I would like to remove #embedded-audio, #gain and #pan from
#audio-speech
... regardless of what happens because they are not required to
support text to speech.
Glenn: I agree because the others can be mixed in.
Nigel: Additional motivation for
doing this is that they cannot all be implemented on the
... same individual content element right now.
github: https://github.com/w3c/ttml2/issues/990
Nigel: Summarising discussion
before we hit this agenda topic,
... Glenn to prepare pull request removing #embedded-audio,
#gain and #pan from #audio-speech
... I'd like to change #embedded-audio to #audio in
#audio-description.
Glenn: I'm happy to do that, especially if it's a barrier to getting the spec out the door.
Nigel: I think it probably will be.
Glenn: That leaves the question
if there will be a demonstration of #embedded-audio
... If you do implement #embedded-audio should we leave them in
#audio-speech?
Nigel: No, still remove them please.
Glenn: I'm okay with that.
... That's changing the #audio-description and #audio-speech
feature.
Nigel: I propose we leave
#embedded-audio in for the time being and I will signal as
soon
... as I know if we will be able to do it.
Glenn: You also proposed removing
#speech which I argued against.
... I prefer to leave it in. I know you suggested signalling it
indirectly through #speak
... which is possible but I don't like the indirection.
Nigel: I think it's more direct.
Glenn: There's a corner case for
transformation processing, does #speak imply that a
... text to speech processor is required for a transformation
processor?
Nigel: I would scope the
requirement for a speech processor to presentation semantics of
#speak.
... Put it this way, we don't have a processor feature for a
font rasteriser, but any presentation
... processor needs one, for visual presentation of text.
Glenn: That's true, but that
feels like a comment for a CR change rather than a change
to
... make now. If this feature does no harm we should leave it
in.
Nigel: That's acceptable, but not
ideal.
... Should we open a pull request now making the changes we've
agreed to, and then
... another to make any other changes needed?
Glenn: Yes I would prefer to do that.
RESOLUTION: @skynavga to change #embedded-audio
to #audio in #audio-description
... @skynavga to remove #embedded-audio, #gain and #pan from
#audio-speech
Glenn: Note the text in 9.3.1
that connects the output of a speech synthesis processor
... to the web audio input.
SUMMARY: If #embedded-audio is unlikely to be implemented, consider removing later; Nigel to inform the group if this is going to be the case by 21st September.
github: https://github.com/w3c/ttml2/pull/999
Glenn: I fixed the pubrules issue so please re-review.
Nigel: Note that this (build) branch is the one the CfC points at.
Glenn: I'd prefer to merge it.
Nigel: Merging it means the ED gives the impression that we have moved to PR already.
Pierre: It's an editorial
decision. Something else to consider is that for substantive
changes
... merge them into master and then merge that back into the
proposed rec branch, so
... the ED is always synced and the PR branch has the
substantive text.
Glenn: I understand but have not
done that practice so far and do not with to change.
... It's not a bad idea.
... The only intent with this pull request is to get to PR, not
to bless the branch as a review
... branch. Since it's already out I can't change that
now.
... People looking at the ED might get confused.
Nigel: I'm concerned with
confusion also, in case someone outside the WG thinks the
spec
... has already been transitioned to PR when they look at the
ED.
... This doesn't really matter, we've done it both ways for
different specs.
IMSC 1.1 Implementation Report
Nigel: At the moment all
additional features that were not in IMSC 1.0.1 but which are
in
... IMSC 1.1 are also introduced as new features into TTML2,
right?
Pierre: Yes
Nigel: Therefore as noted in the
IR there are no additional tests.
... And if we meet the CR Exit Criteria for TTML2 then we also
meet them for IMSC 1.1.
Nigel: We have two open pull
requests, one being the PR prep branch pull request,
... and the other being a tidy-up opened an hour ago.
... Does that tidy-up address all the open issues?
Pierre: Yes, and Nigel's private email to me about line breaks, purely editorial.
Nigel: Thank you.
... That was about words being moved from one line to another
without any change in text.
Pierre: If you could review that
pull request as early as possible that would be great - it
will
... clean up the PR branch.
Nigel: Thank you, I'll certainly
take a look - happy for others too also, of course.
... Anything else on IMSC 1.1?
Pierre: One thing, more an FYI
than a fatal issue. There are two tests that will be used
on
... the TTML2 IR that are not in the TTML2 test repo, for
disparity and luminanceGain.
Glenn: That's not true. If there
are tests in IMSC intended to be used by TTML2 I'm not
aware
... of them. I have tests for both disparity and luminanceGain
in the TTML2 test suite at
... this point. Those are the official ones at this point as
far as I'm concerned.
... A number of the tests in the TTML2 tests repository are
derived from tests that were in
... the IMSC 1.1 test suite and I've marked those in the XML
comments.
Pierre: There are no TTML2 tests
for luminanceGain and disparity in ttml2-tests now. I
... just searched.
Glenn: [looks for them]
Nigel: I found 10 results searching for luminanceGain.
Pierre: There are no presentation tests.
Glenn: That's correct. We should
put some in there just like Nigel submitted for audio,
... if you would like to add them to the presentation tests
then we can put them in there.
Pierre: I sent them in an email to you.
Glenn: I will look for those.
Pierre: Unless they are broken please don't change them and use them as is.
Glenn: I may remove the IMSC profile.
Pierre: Yes, that's fine.
... Can we approve the IMSC 1.1 tests pull request?
github: https://github.com/w3c/imsc-tests/pull/67
Nigel: What's the scope of these?
Pierre: They are intended to
cover TTML2 features that are present in IMSC 1.1 that
were
... not present in IMSC 1.0.1. They are not intended to be CR
exit criteria tests, but tests
... to assist implementers. There are source files and PNG
renders, following exactly the
... same pattern as the IMSC 1 tests.
Nigel: Sounds good to me.
... I've approved them so they can be merged.
... Thanks for those.
... This is a really good resource.
Pierre: It's to help implementers check their implementations are behaving correctly.
Nigel: We said we would publish
after resolving the open issues.
... Let's look at the status...
... We have no open pull requests and the only open issue is
marked for IMSCvNext, i.e.
... later than IMSC 1.1.
... So we can now publish as a WG Note.
... Thierry, what do we have to do to publish as a Note?
Thierry: Just let me know and I can publish it.
Nigel: Okay, we have a resolution, it's completed, please could you publish it as a Note?
Thierry: OK, yes, sure.
Nigel: Thank you!
Nigel: We've covered everything
on our agenda, so thank you everyone, meet same time
... next week. [adjourns meeting]