See also: IRC log
<scribe> agenda: this
Janina reports no complaint to our desire to restore sourcing text tracks of any media type. These were unintentionally removed when WebSRT was removed from the spec. We will have a change proposal to restore this week.
<scribe> scribe: janina
Sean: We seem to be on track coalescing on #4 with the changes we need. Not all points clear yet, though.
<judy> here is the link from the start of silvia's post summarizing status of issue 132
<judy> http://lists.w3.org/Archives/Public/public-html-a11y/2011Apr/0142.html
Bob: I think it's important that
we have simoultaneous video, though understand we may not want
it in a single element.\
... So multiple video tags?
Eric: Definitely possible. Seems
a silly prescription though, if a user agent can support
multiple videos in a single file.
... Doesn't seem important to push just because Safari can.
Bob: So, if multiple video tracks in an inband, it's not clear to me how to display that.
Eric: You cannot do a single element. You need multiple elements and to use media fragments
Sean: We need to be able to discover how video is laid out, and we don't have that yet as far as I know.
Eric: Definitely the case
Sean: Are these always
rectangular? Or can we have a traveling mat?
... If we eventually want to support alpha blending, then we
need to add that to css or adopt what video support has.
Eric: Can do with fragment urls,
but there's a bigger issue. Not all compression support
alpha.
... So not something we can rely on CSS for, as it depends on
compression.
<Sean> Its not an issue then to only have one video per element
<Sean> because transparency and composition of fragments should work
Eric: Looking through last week's minutes, we didn't follow up whether Bob had any issues?
Bob: Don't think so, a question,
though ...
... So a separate video element for multiple tracks, that's
OK
... What about the case where there's not ml and we want to
discover via js
Eric: Not currently at a
computer, but there's an attrib 'exclusive' and believe you'd
use that
... That doesn't provide size or positioning
Sean: Also not sure media fragment can be constructed sufficiently from that data ...
Eric: Using index?
Sean: Not sure. Don't know fragment syntax sufficiently
Eric: You might be right. There may not be sufficient data.
Judy: Can we take up a tangent
for the moment?
... I'm trying to map our recent discussion against various
usre requirements ...
... Thinking of a video with audio content that's captioned,
but also there's a sign translation stream
... Covered, yes?
Sean: Yes, covered. The ml will be multiple elements, but these can come from a single resource.
Judy: OK, thanks.
... I know some people following our discussion, who aren't
directly participating in it, have been confused about
this.
Sean: Yes, that's what I was getting at earlier--the picture in picture overlay for the signing track.
Judy: Also, one other ...
<Sean> Silvia are you joining by phone?
Judy: So my other question, how do we allow for someone who may need text descriptions?
Sean: We think the transcript may
be best for deaf-blind user. But, if needed with the same
timing, then you'd have a text description track and a text
caption track, both enabled.
... Reviewing for Silvia, is there enough data to create media
fragment URI via js
Eric: From single video element
Sean: So, if you want to play two, where you don't know them in advance, is there enough info in the 'exclusive' attribute to create the URI?
Silvia: Checking ...
Eric: Index won't work, because don't know whether there are other tracks
Silvia: For fragment we need the name which is available via audio tracks list or video tracks list
Eric: Name is required? Not every track will have a way to list, nor are they necessarily unique.
Silvia: That's why I was asking for ID
Eric: Both mp4 and qt tracks guaranteed to have unique ID
Sean: And that ID can be used to construct the fragment URI?
Silvia: Yes.
... For slave tracks--which won't have more than one ID
... So a video with multiple audio dubs, we'd need also to look
at getkind and getlang. From these you can decide and do
getid
Eric: Why create different elements for inband audio?
Silvia: Just making something up where more info would be necessary.
Sean: Think one more would be
needed, as elements aren't exclusive. To swithc on/off, would
need to be in a slave element, no?
... Can you switch on a subset? Or is it all or nothing?
Silvia: You can switch on a subset with fragments?
Sean: Mean js
... So, can just switch on/;'off tracks
Silvia: Yes
... Yes also to retrieving data for fragments.
janina and judy expressing happiness at better multilingual support for pwds.
Eric: Any reply on the changes we need to your email, Silvia?
Silvia: Still hoping we get a
positive response.
... Philip's reply had some thing about track list for Eric?
May be what Ian is waiting on.
Eric: Safari won't stand in the way--we want interoperable behavior.
Silvia: Would be good for you to
reply on that.
... What's the opinion on loop and autoplay.
Sean: Autoplay should go away.
Eric: Strongly disagree. If not attrib, then will always be done from js
Frank: Rationale?
Sean: To not bombard the user, if
there were a way to querry ua in script.
... It's dangerous unless ua can reliablbly disable
autoplay.
Silvia: Should be setable in ua a11y settings.
judy: Not sure that disabling it everywhere is the optimal solution.
Sean: Mechanism for switching
play from script would be the same.
... Not sure the suggested solution works.
Silvia: So, step by step
...
... Think it's good to have an autplay attrib as it encourages
its use rather than using js to achieve the same result,
and
... Provides a mechanism for AT/UA to turn the behavior
off.
... So we should also have this on the combined multitrack
element.
... Suggesting the union be the state. So, if one slaves has
autoplay, all slaves will be on autoplay. Though we could do
the opposite?
Sean: Not given this sufficient
thought yet to know the best approach.
... Presumably autoplay is honored only if element is
selected
... Need to think through all the issues.
... Wondering if all shouldn't be the preferred way.
Silvia: Yes, that's the
alternative but more problematical for authoring.
... Autoplay only affects the beginning of presentation.
... Suggests another issue, though, where someone constructs a
js where some autplay, and others don't
... Seems we need to address this thoroughly and resolve
correct behavior.
Sean: Controller hasn't an autoplay; but autoplay on elements sets controller to playing--or not. So the last one creates the actual condition.
SSean: only the element that creates the controller can determin play or not
Silvia: Suggest more discussion on list
Eric: Not sure--need also ml
Sean: This is how I understand the proposal.
Silvia: Yes
... Different behavior for stand alone and combined
Sean: Dangerous
Eric: Just because first element--with autoplay on--insufficient to play if others not sufficiently loaded
Sean: Yes, but that is the decision point
Eric: Yes, we need to hash this out better
Silvia: Reading from spec on if blocked condition ....
[general agreement to add need to control autoplay to Sec 4.1 of our User Requirements]
Silvia: And looping?
Sean: Think it's the same
issue
... Looking at it, I don't think autoplay works as currently
spec'd
Janina: Concerned about Easter related hollidays?
Sean: I'm out Thursday through Monday
Silvia: I'll be working through
<judy> [judy was asking about tonight as well]
Discussion of ready state
Silvia: Also a question if it
exposes too much of inner workings
... Each browser may have different state with the same amount
of loaded data
Eric: Events are not consistent
across browsers
... If a script doesn't register before event is fired, there's
no way to determine whether it should play or not
Silvia: Isn't 'canplay' telling that
[general agreement we want ready state]
Eric: Controller firing events on state very important
Frank Concur
Frank: Similar behavior as possible is a good idea
Sean: No problem in principle, but perhaps in implementation ...
Eric: Minimum value
Sean: Always incrementing?
Eric: Can go backwards
Silvia: But not hard to calculate
minimum
... Simple for browser, not easy for js developer
Sean: And, if you don't have it, you're in worse shape because you have to listen to all events
Eric: Fundamentally, one never knows, but it is much less work and very convenient to have the events and states
Silvia: I don't have additional
issues--nothing new today except autoplay
... Will post, please all contribute to the discussion
Sean: If Ian's proposal not modified by deadline, we still have jultiple proposals
Eric: If we don't agree with his version, whether modified or not, we create a CP against it
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 WARNING: No "Topic:" lines found. Default Present: janina, Sean, Bob_Lund, Eric_Carlson, Judy, Frank, silvia Present: janina Sean Bob_Lund Eric_Carlson Judy Frank silvia Got date from IRC log name: 18 Apr 2011 Guessing minutes URL: http://www.w3.org/2011/04/18-media-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option. WARNING: No "Topic: ..." lines found! Resulting HTML may have an empty (invalid) <ol>...</ol>. Explanation: "Topic: ..." lines are used to indicate the start of new discussion topics or agenda items, such as: <dbooth> Topic: Review of Amy's report[End of scribe.perl diagnostic output]