See also: IRC log
<scribe> scribe: nigel
Nigel: Today, we have the CfC
update, confirming the Implementation Reports,
... and if possible requesting transition to PR,
... IMSC vNext Reqs publication,
... TPAC 2018 planning
... ... anything else?
Pierre: [need to leave in 30 minutes]
Nigel: Ok, anything else for the agenda?
group: [nothing]
Nigel: As I emailed to the
reflector yesterday, the CfCs for the PR versions of TTML1 3ed,
TTML2 and IMSC 1.1 are closed with
... no objections received, so those are now recorded
resolutions, recorded in my emails.
... Thank you everyone for reviewing those documents.
... The only remaining preconditions I'm aware of prior to
requesting transition are to confirm we have completed
the
... Implementation Reports, which we're about to come to.
... Are there any other tasks to do?
Thierry: I've prepared the transition requests.
Nigel: I've reviewed and fed back on the TTML2.
Thierry: Yes, that's done. I
think I'm waiting for a response on the other two transition
requests. I've made a few edits
... so I'll send those to you to review.
Nigel: Thank you, I'll do that.
Thierry: Also all the WBS
questionnaires for AC review are done.
... Once the transition is requested I will need to draft the
announcement.
... We're ready and on time.
Nigel: Just an aside - the
detailed timeline has been an incredibly useful resource for
tracking what we're doing
... and driving us forward, and everyone has worked very hard
to meet those dates. Thank you.
Thierry: Thank you.
TTML1 3rd Edition Implementation Report
Nigel: Note I've created a category for the Implementation Reports:
Implementation Report Category
Nigel: I'd like to do minor
polishing on this to reference the CR Exit Criteria, group the
implementations by organisation,
... and be explicit that the CR Exit Criteria have been
met.
... Aside from that, every test has at least two passing
implementations from at least two organisations, so we're good
there I think.
... Any other comments, is everyone happy with this and my
proposed small changes?
Glenn: Question is "are there any objections". No objection here.
Nigel: Thank you, silence is assent here!
group: [no objections]
Nigel: Thank you, in that case I
will make my editorial changes and we can proceed.
... Do we need a "this is final" banner on the top?
Glenn: Not yet at least, we might have some tweaking as we go further.
Pierre: I think we'll do that once we've published.
Glenn: Exactly, once we go to Rec we can add that banner.
Nigel: Ok that works for me.
Pierre: Just to confirm, on TTML1 we're good to go?
Nigel: Yes. I will review the transition request and then we're good to go.
Pierre: Thank you.
IMSC 1.1 Implementation Report
Nigel: This IR shows no tests needed.
Cyril: I think it's a bit awkward
to have the empty tests section.
... Suggest replacing the 3rd sentence "For this
specification..." to point to the IRs of TTML2 and IMSC
1.0.1
<cyril> The following is the implementation report for the IMSC 1.1 specification. This implementation report is used to demonstrate implementation experience for transition to Proposed Recommendation. All features of IMSC1.1 are implemented and documented either as part of the implementation report of IMSC1.0.1 (link) or as part of TTML2's implementation report (link). Do not hesitate to direct questions and/or results to the TTWG mailing list [1] and/or the IMSC1.1 Ed
Nigel: I like to reproduce the CR
Exit Criteria in the IR for convenience, but I like the other
text.
... I'll also propose changing the "Tests" heading to
"Features"
... I can take those two suggestions editorially and make that
look better.
Cyril: OK
Nigel: Any other points on this?
Glenn: I just fixed an out of date link to the TTML2 IR that was in the IMSC 1.1 IR, FYI.
Nigel: Thank you
Glenn: The TTML2 IR was in the
wrong location, which is why I think the CSS wasn't working
before, because it has to be
... enabled on a per top-level-directory basis.
Nigel: That makes sense.
Glenn: By the way on the transition request do you need the presence of any of the Editors?
Thierry: I don't think there will be a call. If there is a need they will schedule a call, but it shouldn't trigger a call I think.
Glenn: Yes I don't see any reason
why we need to. In the past I've attended when there were
questions.
... I don't think we have any outstanding objections to deal
with here.
Nigel: I'm not aware of any.
Nigel: First, thank you to Glenn
for putting the detail together on this, and for everyone else
for working on the upstream
... online spreadsheet. This has been a massive effort.
Glenn: Spaciba. Thank you.
Nigel: In the past few hours I
have:
... * added the CR exit criteria
... * Added the organisation supplying each
implementation
... (please let me know if you don't want your organisation
listed!)
<glenn> очень хорошо, спасибо
Nigel: * Added a column to Table
3 assessing the CR Exit Criteria.
... They all pass!
... Glenn already asked for people to check their
implementations are correctly represented. If there are any
changes please
... let Glenn know for the totals, and me know in case the CR
Exit Criteria test needs to be changed.
Cyril: It's a lot better, thank
you Glenn for having reorganised it.
... I have one concern that the #T column is misleading to
count so many implementations when we have at least 8.
... I don't think we need the #T column to be so big.
Glenn: It's accurate as defined
and will take effort that's not needed at this point to change
it.
... The data was generated by a formula in the
spreadsheet.
... What you're suggesting would require going through and
changing that.
... It's not misleading if you read the description.
Nigel: Actually they're
misleading in another way because they count non-independent
implementations the same as
... independent ones. That's why I added the CR Exit Criteria
column. The totals aren't actually useful - I didn't use
them
... when assessing the CR exit criteria. It will be a lot of
effort to remove those columns.
Glenn: I wouldn't object to them being removed, I just don't want to do the work!
Cyril: OK I guess I can live with the totals columns remaining.
Nigel: From my perspective this IR is complete. Are there any other changes that anyone wants to make on it?
Glenn: I should note that when I
created the validation test results tables it was a bit of a
manual process because the
... implementers did not provide me data in terms of pass/fail
tests. What I had to do was take the data from the
features
... table and mutate it to create the data for the validation
and presentation test results. So wherever for example a
validation
... implementation had reported F (full) meaning "passed all
the validity and invalidity tests" I could check off all the
tests against
... that feature. Otherwise I did not do so for that
implementation.
... So for those that were marked S I could only mark off the
valid validity tests because I didn't know which of the
invalidity
... tests failed. If you wonder why there are no Fails in any
of those tests, that's because I had no Fail data from any of
the
... implementers. I don't think there's an issue there. That's
why I asked implementers to check their column.
Nigel: That's a good point, do we have the test pass/fail data?
Cyril: For NFLX-V I would have to regenerate that data, I don't have it any more.
Andreas: I think we have for the different features a list of which tests failed.
Nigel: The tests that pass are even more important in a way.
Andreas: Okay. You want documentation of which test files pass and which fail?
Nigel: I think we need to know which tests pass.
Andreas: Perhaps I misunderstood.
Nigel: I think what's needed is
that the validation table entries are checked because Glenn
explained that he derived the
... values in those tables and I took it that there is some
possibility of an error.
Glenn: Especially tables 1 and 2.
Andreas: My understanding is that
the minimum entry point to say something is implemented is to
pass all Valid tests.
... So there should be no false negatives, that would make a
valid file show up as invalid.
... If something is implemented then at least it passes the
valid tests.
... For the invalid tests, there may be some who do not reveal
an invalid message, in which case the invalid test did not
pass
... the test.
... And so the conclusion of Glenn is correct that at least all
the Valid tests passed.
Glenn: I would suggest Andreas
that you have IRT check all of the Table 1 and 2 entries for
subcheck and Cyril the same for
... the NFLX-V column.
... Primarily table 1.
... I just looked again and see that there were a couple of
features that IRT did not report either F or S for animation,
so they
... did not get checked as passed under Table 1 Validation
tests.
Andreas: Ok
Glenn: All the rest were either F or S. I think in the case of NFLX-V all but time-wallclock got reported as F or S.
Cyril: That's correct.
... All of the others were partial or full.
Glenn: I could discern from
looking at the result that the validators that reported that
they passed the invalidity tests were
... doing so for values that were tested by the schema. For
example all attributes that are typed in the schema as an
enumeration
... type were checked off as Full pretty much. That makes sense
to me. The ones that are not typed as enumeration values
... require special purpose parsers which is a little extra
work.
Andreas: I'm just checking. I
think for the Valid tests it is straightforward, it's always
passed if we checked something. For the
... invalid content test to mark it as Passed, it's only for
tests that are implemented fully, correct? Just for F?
Glenn: If you had reported the
feature as Strictly passing (S) then I checked off all of the
tests as passed in the Valid content table, Table 1.
... If neither F nor S was reported for a feature then I did
not check off the tests associated with that feature in the
Invalidity test content table 2.
Nigel: And for F you checked off both of them?
Glenn: That's correct. Sorry that was hard to parse!
Nigel: OK, please do check those columns. Is there any more on that?
Glenn: I'm not expecting any changes unless Table 4 needs changing.
Nigel: Thank you, any other points to raise on the implementation report.
Glenn: This IR for TTML2 was a
lot more complicated than for TTML1.
... And many more tests.
Nigel: Yes, and all the better
for it!
... In that case, I'm declaring that barring any final checks
of table entries, this implementation report is confirmed.
Nigel: The only thing to report on this is that we published the IMSC v1.1 requirements.
Nigel: Thank you Thierry for getting that published, and everyone for contributing to it.
Nigel: I've edited the Wiki page
Nigel: After last week's call I updated the wiki page and agenda to match.
Glenn: I have some input from
some people that they are interested in fleshing out Karaoke
features or at least verifying that
... what we have can support it, or any barriers to usage. If
we need to add any features to support it, along with
live
... captioning I think we should also look again at roll-up
support and see if we can do anything to make it easier to do
that.
... We need some tests that verify we can emulate roll-up using
our current animation technology. I'm in favour of having a
few
... minutes to discuss these items.
Nigel: Ok we have time to do that.
Glenn: I should probably also go through the 2nd Ed issues logged against TTML2 to see if some of those bear discussion.
Nigel: Sounds good.
... We should probably lump 2nd Ed requirements into the
"Future requirements" section.
... I think we should consider celebrating on the Monday
evening.
Glenn: Sure, let's pick a venue.
Nigel: Let's do that offline.
Andreas: Before, there was just
the feature documentation, on the wiki page now we have an
entry for every test file.
... We did not give any information on the individual test
files although we had that information.
... I think all the marks that Glenn set should be okay. The
question is if we should then mark some additional files
as
... passing. The feature may only be strictly passing, so there
were no false negatives, but some features may have
reported
... false positives, but I'm not sure if this is needed at this
point in time because we have enough to cover the
implementation.
... I'm not sure if we should update the table.
Nigel: I agree, the thing we
should avoid is a misrepresentation, so if the page shows a
test passing, but in fact it does not,
... then we need to address that quickly.
Andreas: Yes, but this should not be the case, as Glenn described.
Nigel: The situation we want to
avoid is a misrepresentation - it's particularly concerning if
an implementation claims to
... pass a test but in fact does not. If you want to make
additional claims about tests that pass but are not marked as
passing,
... that's also fine.
Andreas: Okay I understand.
Glenn: We had from last week's
meeting a request to update the informative references to W3C
specifications, specifically
... to CSS documents and to Web Audio. In the intervening
period Pierre pointed out that the URLs for the two XML
Schema
... normative references were not specific URLs even though the
text of those bibliographic entries had dates in them so
they
... could be implied. In both cases those URLs were updated,
and in the case of the informative references in some cases
I
... updated the list of authors, editors etc. It's all up to
date and all the W3C documents referenced now have specific
dated
... URLs in both the normative and informative sections.
Nigel: Thank you for that update.
Nigel: We've completed all of our
agenda topics, thank you. I will review the two remaining draft
transition requests straight
... away and then Thierry when that's done you can make the
requests.
Thierry: I will do that.
Nigel: Thank you everyone. [adjourns meeting]