See also: IRC log
zakim this will be 2119
<scribe> scribe: janina
<JF> zakim: aaaa is Mark Watson
<silvia> zakim: aabb is me
Eric: We're a bit closer.
... Frank hasn't dropped his proposal yet, but I believe he is going to.
... There's quite a bit we like in Ian's proposal and are expecting some modifications from him based on discussion
jf: are all these discussions being captured in W3C space?
Eric: Don't know ...
... I agreed to forward things, if they weren't
Judy: Do we want to specify that the discussion on multitrack be copied to the TF list as well?
JF: Frank, Sean?
judy: just to clarify, it could be both html and tf lists
janina: would prefer it was both lists, it's a critical discussion
silvia: I promissed to write a
summary, apoligize it's not done
... Ian's proposal is specified in what
... I'd like to discuss it.
janina: Is it at least referenced in W3C space?
Silvia: Yes, it's referred in the
bug on this issue
... We have been granted more time for this issue ...
... We still have four fundamentally different proposals on the table
judy: Just concerned that we use the time of people on this call well
Silvia: Would like to put the
controller approach on the table for discussion
... Ian suggests simply using js rather than ml
... Wondering about advantages vs disadvantages of this
Eric: His proposal still
associates the elements in ml
... It explicitly creates a controller object
... Negative aspect--majer negative--that timelines are completely independent
... It would be possible to play these at different rates,; but this could be confusing for users and hard for devs to implement
... So, we at Apple aren't so thrilled with that
Silvia: Can you touch on the good things?
Eric: Yes, we could make this
work more logically
... Some aspects of Ian, ours, and Seans, for example slaving to a common clock
... We gave this feedback to Ian last week--but it was on the what list, sorry
Erioc: Ian did commit to make this change
Silvia: I believe he's working on it, but I don't know the status
jf: Do I understand that Ian's controller approach fits with Sean's approach?
Eric: One aspect of Sean's proposal, last I looked, was to have a text element associated, and Ian doesn't cover that
jf: And to position the text content?
Eric: Yes, we agreed to separate
that last week
... We agree it needs solving, but it's a separate issue
jf: Do we address it before last call?
Silvia: At the moment we can render text anywhere on page manually. We want to improve on that using css selector, and Sean created a wiki for this
Sean: Yes, there are several
issues around the api for selecting tracks, but they can be
... There are good things in Ian's proposal
Eric: It would be easier to solve text rendering apart from Issue-152
Sean: I don't think Ian's has a clean enough approach on mutually exclusive
Eric: Agree that requiring inband video tracks to be mutually exclusive doesn't make much sense
... Also don't understand why audio can be multiple, but video is exclusive
Silvia: I believe we can
... Frank, have you looked at Ian's Thoughts?
Frank: Agree there much merit. Agree we start from it
jf: Does anyone disagree?
Eric: We need to ask when we can
expect an updated version from Ian
... It would be helpful for us to continue from an updated spec that incorporates our ideas
Silvia: I'm messaging him ...
JF: Main concern is that we have to keep momentum on this
Silvia: I have an additional
question re controller
... Inessence this means defining an abstract timeline apart from each of the component elements
... Is that possible? An abstract timeline or do we need to pick one?
Eric: No problem because that's
what will happen in the background anyway. All elements will be
driven by the hw clock. It's an implementation detail
... At least in our implementation
Frank: Can you say more about time offset zero?
Eric: Having time 0 correspond to
one of the files won't be that hard
... From the user's perspective don't think it would make sense to have nothing rendered for awahile
Silvia: Time 0 is where they all start?
Eric: It may make sense to specify a start time other than 0.
janina: What about a use case where the alternative is provided in multiple files?
Eric: Don't doubt that this is the case, but it isn't hard to build the composite file
Silvia: Don't think multitrack is
the best answer for that
... Agree that flattening the individual files into a single file is easier for publication, use, etc
Mark: We currently store each subtitle, etc., in a separate track
Silvia: But not segmented in time?
... Each file contains the entire duration
janina: So, continuing as the devil's advocate, what about extended video descriptions?
Silvia: Yes, that changes the
... It's a different problem, and one of the most complex
... Would not to like to solve it in multitrack
Eric: But segmented files wouldn't solve the problem either, because all others would need to be paused while this one continues to play
Frank: Prefer a simpler approach,
offset seems too complex
... Agree that linking multiple segmented files into a single file should be done in authoring
<silvia> …authoring the media
jf: So thinking about sign language ... if the sign language tracks stalls for some reason, it should tell the main to pause at some point
Frank: Yes, we can do that today
Silvia: Am I understanding that the controller isn't the best idea? That one element should be the primary timeline?
Silvia: So can't you achieve that in implementation? Just pick the one you want to use?
jf: My concern there is authoring; Now we have to tell people the first video is primary and may not be correct oftimes
Eric: Silvia's suggesting that
that one timeline could be chosen to use as the master
... It's only necessary to expose this to authors if we don't require because of data load problems
Silvia: If we allow any element to be the master, users might prefer different choices
Frank: if two elements have different length content, how do we reconcile
Eric: Propose that the duration
of the group is the union of its members
... Ian has pointed out that it's sometimes not possible to predict
... So should be as in video editor, the longer file extends the shorter
Silvia: So the longest rules
Frank: And the use case?
Eric: A movie with director's comments often go longer
Mark: There's also different packet lengths between audio and video
<mark> differentpacket durations
Eric: So if you enable an element
that had been disabled, it may make the composite presentation
... Also think we should apply this principle for display
Silvia: How so?
... They're already displayed in different areas, no?
Eric: Maybe, maybe not
... I suppose depends on controller and what properties we have there
... If we require all elements for some time be loaded before we play,
Silvia: So an abstract state from all the elements that are active?
Eric: This was in our feedback to
... Restating for Sean who some of us couldn't hear ....
... Sean smil has this same issue and adopting the behavior that duration of group can be longer of any element we need to decide what happens with elements that have video when play goes beyond the end of video
... Suggest rendering nothing
... Suspect you wouldn't want last sign gesture to be held
jf: what does smil do?
Eric: allows to be specified
<Sean> good summary
Silvia: Think going to transparent
jf: so that box would disappear?
<Sean> does it still contribute to the overall shape however?
Eric and Silvia say 'yes'
Mark: Isn't the usual behavior to freez
Eric: last frame is held because time stops so you see the image for the last frame
Silvia: Some display links to other videos
Mark: And if the track is disabled?
Eric: Then it disappears from rendering
Frank: I am concerned whether we're taking on more complexity again
Eric: How different?
Frank: If longest element wins, what do you do with a loop event?
Eric: And how is it simpler in the other case?
<Sean> Is it always the case then that effectively the longest video is always the master?
Frank: Because we've defined the master
Silvia: It's the case for both approaches
Frank: Don't agree
Eric: So, if we don't require
slaved elements ... if we don't pause the timeline because
every sample isn't loaded ...
... If we require all must load before we play, then I agree it's the same from the implementation viewpoint
Mark: Sounds like it's simpler to have one master, whichever is the master
Eric: Don't agree, but let's
... What do you do with the longer slave?
Mark: It lengthens the
... So the master is extended to the longest of any of the active elements ...
Silvia: Yes, and that's the
... So we change much data about individual elements when we play a group -- so it makes sense to consider the group as the master
Frank: I'll reconsider the proposal
jf: Four minute warning ...
... Silvia, some work on wiki, yes?
JF: We'll ask Ian to cc W3C lists
judy: Would it help to have
Silvia's summary list remaining questions/issues?
... Can we collect now?
Eric: Several: What should ---
Should all group elements have the same controls?
... Loaded ranges?
... What should happen when script does a seek on one element
... Ditto for other attributes -- What happens?
<Sean> Should in fact media elements have any means of controlling other than the controller
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Found Scribe: janina Inferring ScribeNick: janina Default Present: John_Foliot, Judy, [Microsoft], Eric_Carlson, +1.408.540.aaaa, +61.2.801.2.aabb, silvia, mark Present: John_Foliot Judy [Microsoft] Eric_Carlson +1.408.540.aaaa +61.2.801.2.aabb silvia mark WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Got date from IRC log name: 06 Apr 2011 Guessing minutes URL: http://www.w3.org/2011/04/06-html-a11y-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]