See also: IRC log
<JF> ah, cool, we can use a11y?
<janina> Looks like!
<JF> as in a11y# ?
<JF> dialing in now...
<janina> Waiting for you!
<janina> Scribe: Janina_Sajka
<janina> agenda: this
<janina> Chaas, are you dialing in? We're on Zakim 2119#
<chaals> thanks.
<chaals> I joined to find out.
<janina> There's a new CP on the floor, though!
<JF> http://www.w3.org/html/wg/wiki/ChangeProposal/Issue194_SP
<JF> just in case chaals hasn't seen it
<janina> Of course, if Silvia doesn't wake up and grab some coffee, this may b a short meeting after all!
<silvia> Hi everyone
<silvia> sorry, I just found the channel
<janina> Hi, Silvia, code is 2119#
<JF> dial in code is a11y#
<janina> Or you can SIP 002119@voip.w3.org
<silvia> phone number?
<janina> U.S. +1.617.761.6200
<silvia> thanks
<janina> scribe: janina
jf: Background from the F2F just
concluded
... Discussion with Ted started over dinner
... I thanked him for his research paper
sp: I helped him a bit with
that
... He wrote it all up
jf: So thanks also to you!
sp: I'm still of the mind that we haven't thought it all through sufficiently yet
jf: That's also the critical
thing for me. We shouldn't be deferring this to .next
... Not so concerned whether we get it over first LC or second
LC, but we need it in this spec
sp: I'm most fussed over getting it right and not so much as whether it's this rev or .next
chls: I'd take a middle path, we need to go the right direction ...
<JF> JS: the more I have been thinking about this, less worried about the mechanism and more about user-accomodation perspective
<JF> becoming concerned about a unified presentation of options to the end user... button creep
<JF> if we don't have a mechanism that allows for reliable grouping, we could end up with too many controls in the UI
<scribe> scribe: janina
<JF> ach ch
chls: I'm worked about having a
reasonable uniform mechanism
... If we do for real come to a button creep problem, we'll
have to deal with it
... But we probably won't get it right the first time
... The authoring complexity is my concern with the point-to
link
... Strikes me much of this discussion is a proxy for
longdesc
... The cable participants thought there was something else
that was a hidden agenda when we talked about this during the
F2F, which kept them out of the discussion
... I heard that in a side conversation
<JF> +1 to the perception that the initial discussion at the F2F was an @longdesc proxy discussion
chls: There won't be that many really full presentations--except large movie sites, government,
janina: Also education
sp: Which proposal are you referring to?
chls: The design pattern of pointing to a link external to the video element, as opposed to something inside it
jf: I would echo that as
well
... The proposal that Ted was actioned to write would have an
idref pointing to a uri
... As we discussed this more and more, many of us, especially
a11y, came to like the tracks proposal best
... seemed simple and self-contained
sp: Superficially it's clearly
that attractive
... Because it can be any format, pdf, odf, docx, daisy, it's
more like a longdesc
... Is supposed to be a defined format
... Files that we want renedered on page, as well as those that
we might buttons for, etc
... this makes my mind explode as to how to author, how to
develop the implementation, etc
... There's no extra screen region to refer to outside the
video element
... That's why i became so unhappy when I saw this
... So I backed off to basics ...
... The transcript doesn't require the video, you can consume
it on browsers that can't render video/audio
... That's how I came up with the element outside the video
element
... How we render, how we do the ui, we still need to work it
all out
jf: My concern is still the copy and paste part
sp: but that's not how people embed videos
jf: Tip of the hat to youtube for
figuring out a good way to do that
... But will that be the only model?
... Just concerned that the authoring pattern be very clear
sp: I appreciate the concern, but
as a well developer, when I copy functionality to another page,
I simply make sure I get it all
... All the sites I've seen that publish video provide a way,
either iframe or a code segment that is selected and
copied
... even now it includes more than the video element, title for
instance
chls: ... looking at the tracks
proposal
... Not sure how transcript in tracks, with 0 time ... how is
that handled
... rendering into video space is probably not such a big
issue, because people who can't see the video won't care about
rendered video
... also not so concerned about linking -- we know how to
handle links
... do exactly what Opera did with longdesc -- that does seem
to work
... Our experience is that it works tolerably well
sp: At least in webvtt at the
moment, we require start and end, so if it's only 0 it will sit
there in a single queue
... js will have to do something with it--no rendering
implied
... But then we restrict ourselves to transcript either in
webvtt or ttml, but not the other file formats
... It doesn't give us rendering, what do we expect the web
browser to do with it?
<JF> +1 to overt restrictions on trnscript file formats
chls: Agree that it restricts us
and makes a mess because a lot of docs won't be available in
that format
... we have the kind attrib to define what to do
<JF> <foobar kind="transcript">
chls: It could link out to a doc, pdf, etc
jf: Also a web developer--how I pay my bills--and work with a building full of other developers
chls: These still do silly things, hence my concern. We need to make the right thing simple and clear
jf: That's why I'm looking at the authoring pattern, not so much the underlying mechanism
sp: How then is the transcript rendered when it's inside the video element?
jf: via js, i suppose?
... Will work one way on my phone and differently on my
laptop
sp: video element, if we put
stuff in there, affects a defined region of the screen.
... In recent years people have been backing off from too many
buttons inside video
... what seems to be of issue is that we're only talking, so
far, of one way of dealing with transcripts, whereas there are
two other ways as well
... We have the not on screen but programatically discoverable
transcript ...
... everything inside video is passed away by the browser--it's
not in the dom
<JF> <video id=v1 src=video.mp4></video>
<JF> <p>Lorum Ipsum X 15 paragraphs</p>
<JF> <transcript for=v1 id=t1>
<JF> <a href=transcript.html>Transcript for the video</a>
<JF> </transcript>
sp: we have all the techniques we need for associating transcript and video, even when it's not inside the element
jf: still concerned about the association between video and transcript when it's outside the element
sp: same for labels
... it's only that it makes sense for them to be next to each
other
chls: I'm not so sure that the
hidden discussion from i204 is unrelated to this
... we seem to be building something fragile -- when dealing
with hiding stuff
... relying on a link to content that's somewhere else is the
fragile part that's still problematic -- longdesc like
cls: people are still shipping display=none and 1pixel gifs with alt, etc
chls: to my certain knowledge,
WAI has been working on solving this for 15 years
... I think that an approach which relies on techniques to make
hidden text accessible is a very fragile one
We'll wait for you!
Silvia, from Zakim?
<silvia> yes - it's restricted
<silvia> call center is restricted at this time
<chaals> call center?
<chaals> not conference?
<silvia> the pass code is not valid
<chaals> i.e. is zakim telling you this?
<chaals> aha
Silvia, use 26631#
<silvia> pk
zkim, ??P3 is Janina
sp: q about fragility? why if in video element? don't we have the same problem however we do it?
chls: in the elemtn we rely on
browsers not doing stupid things
... outside, we rely on authors not doing stupid things
... experience suggests that even a million authors doing this
right, there will still be masses of authors doing this
wrong
sp: we still want rendering on the page, no?
[agreement in the room]
jf: I suspect we're thinking the
link to transcript would be displayed as a control
... Citing experience with dlink which has not been
successful
sp: why are we repeating dlink?
jf: that is what relying on a text link is doing
[process discussion on next steps in view that this was to be done May 11 according to F2F]
jf: any reason transcript element can't be child of video?
sp: yes, because it restricts us
to the screen space assigned to video
... my approach gives us anothr view port
chls: agree with concern on
rendering model
... one use case video not even rendered
janina: pointing out that the no video use case isn't just disability use case
chls: side by side still sensible rendering
chls; if has timing, no problem
chls: still not convinced that
it's too hard to mix this in and that it requires external
definition
... e.g. pasting a word doc into an iframe rendered side by
side with the video
... of course it's more readable if read without the timing
ml
... but harder to render in sync
sp; we should handle all types of transcripts in a uniform manner
jf: don't think we're that far apart
janina +1 to not so far apart
jf: still fighting to preserve simple and clear author experience
<chaals> sp: my concern is that you wouldn't be able to render the transcript in parallel with the video
sp: my proposal is similar to the signed translation being in a separate video and the video description in a separate audio element
<chaals> jf: use case is to save the transcript, and read later.
<chaals> [I agree that it should be possible to render side-by-side. I don't see how any of the proposals get away from "the web developer has to make that feasible" (on the basis that browsers getting into hcking the display of pages is a pretty fragile way to work)]
<chaals> [I am concerned about the issue of whether the transcript data should be inside or outside, but I may be convinced that having it outside is workable (I would have to "evolve my position" to get there...)]
<silvia> we're considering another call tomorrow same time
<silvia> does that work for you?
<JF> chaals, we are going to try and reconvene at the same time tomorrow
<chaals> [I am really concerned about building on top of a bit of page content, because I believe that is repeating the D-link experiment. People *will* try to hide the text (which means you lose any benefit over the original longdesc approach), and will get it wrong...]
<silvia> I'll think this through for tomorrow...
<chaals> I'll try for tomorrow.
<chaals> [I am not that fussed about the colour of the bike shed. There are many ways to make a colour scheme consistent in my bike shed]
<chaals> [Oh. I mean that there are multiple ways to get consistency while tossing around attribute/element, special name or re-use something, ...]
rrsagent make minutes
This is scribe.perl Revision: 1.136 of Date: 2011/05/12 12:01:43 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/@chhals/chaals/ Succeeded: s/As I read SP's current proposal, it seems what Ted was actioned to write at the F2F/The design pattern of pointing to a link external to the video element, as opposed to something inside it/ Succeeded: s/pcsl/pixel/ Succeeded: s/not yet convinced that any particular approach proposed is more or less fragile than another/I think that an approach which relies on techniques to make hidden text accessible is a very fragile one/ Succeeded: s/things right, there will be still others doing it/this right, there will still be masses of authors doing this/ Succeeded: s/are we/why are we/ Succeeded: s/it seems so/that is what relying on a text link is doing/ Found Scribe: Janina_Sajka Found Scribe: janina Inferring ScribeNick: janina Found Scribe: janina Inferring ScribeNick: janina Scribes: Janina_Sajka, janina WARNING: Replacing list of attendees. Old list: Janina JF chaals +61.2.801.2.aaaa silvia New list: JF silvia chaals janina Default Present: JF, silvia, chaals, janina Present: JF silvia chaals janina Got date from IRC log name: 10 May 2012 Guessing minutes URL: http://www.w3.org/2012/05/10-media-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]