See also: IRC log
<trackbot> Date: 15 May 2014
<scribe> scribeNick: nigel
nigel: Can't cover this topic fully today as we need Glenn and David to contribute to the conversation, but we can summarise the state of the debate to date.
mike: I'm concerned that we
haven't addressed my email that described the points of
disagreement.
... the thread seems to have evolved towards document
conformance profiles, i.e. documents conform to all of the
profiles that are signalled. Then a decoder only needs to know
that it can support one of the listed profiles
... in order to proceed. This is useful but not sufficient. The
proposal requires registration of the profile, which is fine,
but it doesn't allow 3rd party namespaces to be added and
signalled like the current
... 14496 does. That forces the industry to register all
separate combinations of namespaces. That's not the end of the
world but the discussion hasn't solved that.
nigel: It's a point of dispute in the reflector whether namespace signalling is needed at all - they're features of specifications already.
mike: It would be a problem for Ultraviolet if all the features needed for processing a document can't be defined, e.g. TTML subset plus image processing subset.
pal: Why can't we delegate that combinatorial logic to the registration?
mike: Yes that will work for the
profiles defines by Ultraviolet. But it does not solve the
problem when the document contains 608 and 708 data. There's no
way to signal that. Someone would have to register
... the Ultraviolet profile that includes 608 or the one that
includes 708 or the one that includes both.
pal: When you say 608 is that a separate specification?
mike: yes.
pal: So another row in the specification table could contain m608 and reference that other specification.
mike: Yes. It forces anyone using
a combination of namespaces to register with W3C the
combination. There's no concept in the current proposal to say
'processor needs both X and Y'.
... The proposal is necessary but not sufficient - it would say
'doc conforms to X, Y and Z so if you support any of those you
can decode the document'. This doesn't cover other
namespaces
... which would have to be combined and the combination
registered with W3C.
pal: So you'd like a way to signal 'document conforms to A+B or C or D' to create on the fly new combinations of conformance by combination.
mike; Right. The argument on the thread was that this allows custom profiles to be created on the fly. Today that's already possible in the ISO specs. We were only really dealing with the unwieldiness of the long strings.
scribe: It's not the end of the world as long as the registration process is lightweight.
nigel: It would be really helpful Mike if you could point the discussion to where the requirement for combinations of profiles comes from in the ISO specs.
mike: It would also be helpful to talk about the registration process. If the bar is really low then it's much more compelling to force people to register the combinatorial profiles they want.
nigel: Describes lightweight process.
mike: Maybe we could codify reasons why a registration may be rejected, e.g. Not TTML, Already Registered etc.
nigel: It's for me and David to work on the agenda. The goal is to work on our deliverable of the TTML <--> WebVTT mapping.
issue-307?
<trackbot> issue-307 -- Conformance language and processor profile rather than content profile. -- open
<trackbot> http://www.w3.org/AudioVideo/TT/tracker/issues/307
issue-308?
<trackbot> issue-308 -- It is unclear how a document is associated with a related video object -- open
<trackbot> http://www.w3.org/AudioVideo/TT/tracker/issues/308
pal: We're awaiting glenn's input
on issue-307.
... I think nigel wanted to think more about 308.
... my action item now is to fix the typo re Related Media
Object.
... I believe the current document does not mandate that a
document be attached to a single frame rate.
nigel: okay I'll review more.
pal: I propose to make the change to Related Video Object either by defining it or changing it. Then I'll close the issue and if there's an issue related to frame rate we can open a new ticket.
mike: Historically we made it 'media object' to be more generic than video and the link is external to TTML.
pal: my plan is to define Related Video Object as the video component of the Related Media Object.
nigel: We need to clarify that frameRate in an IMSC document does not need to be identical to the frame rate of the related media object.
pal: Would a note be sufficient?
nigel: yes, possibly.
https://www.w3.org/wiki/TTML/changeProposal002
nigel: I've got a feeling we may
need Glenn to review this before we can close it.
... In fact Glenn has put a note at the bottom showing he's
implemented it so it's for jdsmith to verify that the
implementation meets the expectation.
https://dvcs.w3.org/hg/ttml/rev/c9fd49837446
nigel: that's the change set on
TTML2
... The editor's draft for TTML2 is at
https://dvcs.w3.org/hg/ttml/raw-file/default/ttml2/spec/ttml2.html
jdsmith: I have a question about
change proposal 5. It's not really a change, but a
representative way to map to HTML.
... We've reviewed that here, and it's pretty much what we do.
What's the destination for the change?
nigel: It can be an appendix to
TTML2, or striped all the way through TTML2 per feature, or a
separate note altogether.
... If the whole world is moving away from XSL-FO towards CSS
then it would make sense to add it throughout TTML2 on a
feature by feature basis.
jdsmith: I could map the rendering to the sections in TTML2.
plh: I think this would be extremely useful.
pal: If it applies equally both
to TTML1 and TTML2 it should not be an appendix to TTML2.
... In that case a separate document would be useful.
plh: I can see it either way. You can use it for TTML1 additionally to TTML2.
nigel: I would prefer it to be striped through TTML2 to encourage new implementers to start there not at TTML1.
jdsmith: I'll start working on the mapping to sections.
plh: We don't want to overload
the editor.
... You could potentially become another editor for this
purpose.
... By the way some features may not be mappable if CSS doesn't
have the features, and it's important to highlight them. That
will probably give rise to a sub-profile of TTML2 that avoids
non-CSS features.
jdsmith: Let me look at the work of striping it through, see what Glenn thinks and hopefully confirm next week.
<scribe> ACTION: nigel to create a wiki page for the September F2F with a draft agenda and an 'intention to attend' section. [recorded in http://www.w3.org/2014/05/15-tt-minutes.html#action01]
<trackbot> Created ACTION-289 - Create a wiki page for the september f2f with a draft agenda and an 'intention to attend' section. [on Nigel Megitt - due 2014-05-22].
plh: We're also looking at
setting up an Extensible Web Summit in Berlin in September just
beforehand. This would give people a good reason to be in
Europe around that time.
... The dates aren't final yet - our German office is looking
for a host. The location will be Berlin because on the 13/14th
Sep there's a Javascript conference. Once it's settled I'll
report that back here.
... It may help people to attend.
... On the testing front, there are lots of good movements
there. We have a test suite for WebVTT - I'll add the links to
it to our wiki dashboard. At some point we need to look at what
we can do for TTML.
... The old tests are still valid but we may need to refactor
them so they can be run automatically.
nigel: Change Proposal 26 is our home for this https://www.w3.org/wiki/TTML/changeProposal026
plh: Waiting until September before diving into the tests will be really helpful. Any progress with scripting languages and declarative languages that help automatic testing is good for us.
Re Codecs we're tentatively scheduled for the 29th.
plh: regrets for the next 2 weeks.
This is scribe.perl Revision: 1.138 of Date: 2013-04-25 13:59:11 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/both/equally both/ Found ScribeNick: nigel Inferring Scribes: nigel Default Present: pal, nigel, Mike, [Microsoft], Plh Present: nigel._pal Mike jdsmith Regrets: Frans_de_Jong Glenn_Adams Found Date: 15 May 2014 Guessing minutes URL: http://www.w3.org/2014/05/15-tt-minutes.html People with action items: nigel[End of scribe.perl diagnostic output]