See also: IRC log
<scribe> scribeNick: nigel
nigel: I've scheduled today for
an hour and a half for those who can make it.
... For today, the main topic will be TTML2. There are also
some open issues on the TTML Profiles document, and on IMSC
which we should cover.
... There has been some activity on WebVTT but as far as I'm
aware we aren't going to
... spend any time discussing that today.
... I'd like to think about a next F2F meeting too and check
diaries.
... Any other business to cover?
group: No other business.
group: [discusses dates] w/c 23rd January is a possibility.
nigel: At that time I would hope
we are ready to discuss TTML2 CR exit criteria and transition
to CR.
... Also to scope the next IMSC.
Andreas: EBU Production Technology Seminar is 24-26 Jan in Geneva.
glenn: That week is not great for me - how would Jan 12/13 work?
group: That works ok.
nigel: Okay I'll send a note out to the group including expected location.
<scribe> ACTION: nigel Propose 12 and 13 Jan to TTWG for a F2F meeting in London [recorded in http://www.w3.org/2016/10/13-tt-minutes.html#action01]
<trackbot> Created ACTION-482 - Propose 12 and 13 jan to ttwg for a f2f meeting in london [on Nigel Megitt - due 2016-10-20].
tmichel: There has been some
activity by David and communication with a CSS WG and a
... long email from David that I have not reviewed yet. I also
noted that there are a lot of
... substantive changes so a new WD is needed and a new wide
review. We should wait
... for the time being until David, Simon et al can attend.
nigel: Ok thanks for the update.
nigel: I opened a pull request https://github.com/w3c/tt-profile-registry/pull/28
tmichel: I also raised an issue for wrong URLs: https://github.com/w3c/tt-profile-registry/issues/29
mike: None of the updates I think
are critical. It was important to get the published
version
... out with the updated RFC link in. I'd like to set a
deadline for further comments and
... deal with them. Maybe mid-November.
nigel: Okay that works, about 4
weeks from now. Do you want to leave the current stuff
... as is for the time being?
mike: Yes, I can sit down and gather together all the pulls etc in one go.
tmichel: For information, I have
set up the echidna automatic publishing tool for this
... document so now it is very easy to publish, I can do it
myself. The only thing I need is
... a WG Resolution in the minutes to publish.
nigel: Ok ask us for that when we
have a new version in a month or so please.
... Anything to discuss right now?
mike: They seem editorial to now, the pull requests.
nigel: I think so.
nigel: There is an open pull
request for refactoring the repo, and there's the topic
of
... moving to gh-pages.
Pierre: I don't think there are any problems with the pull request so I plan to merge it.
nigel: https://github.com/w3c/imsc/issues/187
tmichel: Philippe recommends that we move to gh-pages. We have been doing this across W3C.
glenn: I remember reading about
the use of gh-pages and the auto-publishing system so
... that documentation should be available.
tmichel: Either we keep master or gh-pages but both is crazy.
glenn: I agree, we should not keep both. I have seen no problem with using gh-pages for TTML2 so far.
andreas: Also note that the TTML WebVTT mapping document uses gh-pages.
<glenn> https://w3c.github.io/faq.html
nigel: There's no reason not to
have a root level index.html page that provides an intro
... to IMSC and has pointers to spec/imsc1.html or
spec/imsc2.html for example.
Pierre: I would be happy to create such an index page.
glenn: You can choose which
version of the head syncs with the gh-pages branch for
... auto publishing - I suggest reading the faq I posted
above.
Pierre: I will do that.
nigel: First, horizontal review
action-480?
<trackbot> action-480 -- Thierry Michel to Request schedule time for horizontal review of ttml2 -- due 2016-09-26 -- OPEN
<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/480
nigel: One of these things is the
security questionnaire. Thanks Thierry for updating our draft
responses.
... My understanding of "valuable data" is that it refers to
something that can unlock
... greater sources of data, such as credentials, or something
that would allow authentication
... as a particular user. I do not think this affects us at
all.
tmichel: That's as I understand it too.
nigel: The next thing is 3.6 regarding if <set> is a script.
mike: If TTML like animation in
SMIL etc is not considered a security risk we should follow
that.
... The concern is that for everything else in TTML you can't
cause harm, but if you create
... a maliciously crafted <set> you could make a
processor go off into a big loop and
... deny the service.
glenn: Is that any different from creating a lot of content that changes at millisecond boundaries for example?
mike: No, but it may be easier to detect.
glenn: The HRM in IMSC 1 is one way to handle this - does it handle <set>?
Pierre: It does address <set>
glenn: That's one way to do this,
and we could do something similar in TTML now that we
... have an example of this formalism.
Pierre: As I understand it in
TTML1 <set> is an authorial shorthand to create new ISDs,
so
... since the HRM works on ISDs whether the ISDs come from
<set>s or timings or captions
... it makes no difference to the HRM.
nigel: +1
glenn: That's true. When we have
the <animate> element it creates intra-ISD
animation.
... One way to deal with <animate> is to quantise to some
degree at which point you
... will have effectively generated a bunch of ISDs at some
level.
nigel: One thing we do not have
is something like the HTML 'time marches on' algorithm,
... so processors need to be constructed well but badly made
ones or those that cannot
... deal well with the input documents could generate
non-compliant behaviour.
glenn: It would be good to look
up the SMIL and SVG animation language and see if they
... have any impact on security.
tmichel: I can go and look at that.
nigel: Thank you
... Mike spotted that there's a missing answer to 3.8, which is
"No".
mike: The Media Type
Registration's security considerations section are an integral
part of
... the spec even if they are not inline.
nigel: I assume there will be a normative reference from TTML2 to the TTML Profiles registration document, if there is not already.
mike: I would think that is
essential.
... We may not be able to reference a Note normatively.
... We could refer to the IANA document.
glenn: I will insist on using the
same media registration.
... This is a valid question and something we should address in
the content.
... One of the few additions I have made to the minimum profile
requirements is that all
... processors of TTML2 must support the ttp:version attribute
which would be marked as
... 2 or greater if it is not TTML1 so there is an internal
mechanism for noting the version.
mike: That indirectly solves the TTML1 issue too.
glenn: Yes, it is required.
nigel: So Thierry are you happy to update 3.16 to say "Yes"? I see that http://w3c.github.io/ttml2/spec/ttml2.html#media-types-registration shows an issue to add the registration in.
glenn: I think that has been
overtaken by events somewhat so I want to avoid defining a
new
... media type registration if possible and refer to something
external.
nigel: Can we normatively reference the IANA registration?
mike: That seems the simplest thing to me.
glenn: I agree.
andreas: [has to leave]
nigel: The TAG has begun looking
at TTML2 also at https://github.com/w3ctag/spec-reviews/issues/138
... They have begun to look at the spec in ED form and have
asked some questions. I
... have added a quick response already.
... By the way they do ask about the high level summary of
differences, which I have the action to draft.
action-481?
<trackbot> action-481 -- Glenn Adams to Provide nigel with a list of new features in ttml2 to begin reviewing -- due 2016-09-26 -- OPEN
<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/481
nigel: I haven't managed to do
this yet.
... But will do so tomorrow hopefully.
glenn: We can make reference to
comments and responses in the TTML1 process if we get
... pushback from CSSWG, e.g. why we change from hyphenated
names to camel case.
... And why we introduce new value vocabulary relative to CSS.
Ultimately they point back
... to our requirement for interoperability between content and
feature requests from users.
... It would not hurt to provide guidance for targeting CSS
renderers and those presentation
... semantics that may be missing.
Pierre: Do you have a list of
those? A lot of TTML content will end up on the web and
be
... rendered using HTML.
glenn: I don't have an up to date
one. There was some preliminary work generated by Sean
... that could probably be taken from the change proposals. It
would be out of date and need
... updating. I can see putting an informative annex into our
spec to cover this material.
... I think that would be a good idea.
pierre: https://www.w3.org/wiki/TTML/changeProposal005 ?
nigel: Yes.
glenn: But we should not be limited by CSS.
pierre: I think that the use cases go beyond distribution is a compelling argument. Thanks for that.
nigel: Anything more on horizontal review?
tmichel: No, nothing more. I think the other groups are waiting for the WD on /TR.
glenn: On this topic, since we
have some existence proof for SVG renderers, even when
CSS
... does not directly support a feature we can point to SVG to
support that rendering.
nigel: Okay that's all for
horizontal review, let's move onto pull requests.
... https://github.com/w3c/ttml2/pull/197
glenn: I did address your point
about should/must nigel. The other point about origin
... being optional, since origin is optional on region, it
seems worthwhile making it optional
... on safe crop area also.
... In my experience a generic use of this kind of thing is
centered. I do not view the
... complexity of making it optional to be a barrier. You had a
question on some of the
... computation I had put in place. I think you made one
misinterpretation.
nigel: Ah, ok
glenn: The last point is designed
to tackle the case that the safe crop area extends
outside
... the root container region, and pulls it in. I agree it
would have been possible to specify
... that percentages are less than 100%, however you still have
to handle the case where
... the extent goes past 100%. The only syntactic limitation is
to be non-negative. That
... algorithm will always put it inside the region.
https://github.com/w3c/ttml2/pull/197/files#diff-5ac39ab39265717d36dfd789d254ce1aR4932
pierre: I notice you did not take into account my comments about the name.
glenn: I think safeCropArea is easier to understand and I agree with Nigel on this one.
pierre: This is not a permission to crop.
glenn: No it's a prohibition from cropping within the area.
nigel: My other query is if we need higher level protocol or can just use document processing context.
glenn: I think there will be
other places where we use it so I would like to use it here, if
it turns out that there are no other places then I can remove
it.
... I will make the minor tweaks to remove "to" and change the
should into a deferment to
... the normative semantic.
... Then I will merge this, since there are no objections to
doing so.
nigel: Okay we're out of time. Thanks everyone. [adjourns meeting]