Another abandoned server code base... this is kind of an ancestor of taskrambler.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

478 lines
17 KiB

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html lang='en' xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<meta name="generator" content=
"HTML Tidy for Linux (vers 6 November 2007), see www.w3.org" />
<title>Media Fragments Working Group Teleconference -- 24 Nov
2010</title>
<link type="text/css" rel="STYLESHEET" href=
"http://www.w3.org/StyleSheets/base.css" />
<link type="text/css" rel="STYLESHEET" href=
"http://www.w3.org/StyleSheets/public.css" />
<link type="text/css" rel="STYLESHEET" href=
"http://www.w3.org/2004/02/minutes-style.css" />
<meta content="Media Fragments Working Group Teleconference"
name="Title" />
<meta content="text/html; charset=utf-8" http-equiv=
"Content-Type" />
</head>
<body>
<p><a href="http://www.w3.org/"><img src=
"http://www.w3.org/Icons/w3c_home" alt="W3C" border="0" height=
"48" width="72" /></a></p>
<h1>Media Fragments Working Group Teleconference</h1>
<h2>24 Nov 2010</h2>
<p><a href=
'http://lists.w3.org/Archives/Public/public-media-fragment/2010Nov/0033.html'>
Agenda</a></p>
<p>See also: <a href=
"http://www.w3.org/2010/11/24-mediafrag-irc">IRC log</a></p>
<h2><a name="attendees" id="attendees">Attendees</a></h2>
<div class="intro">
<dl>
<dt>Present</dt>
<dd>Yves, Raphael, Thomas, Silvia, Philip_(irc)</dd>
<dt>Regrets</dt>
<dd>davy, eric</dd>
<dt>Chair</dt>
<dd>Raphael</dd>
<dt>Scribe</dt>
<dd>Raphael</dd>
</dl>
</div>
<h2>Contents</h2>
<ul>
<li>
<a href="#agenda">Topics</a>
<ol>
<li><a href="#item01">1. Admin</a></li>
<li><a href="#item02">Discuss HTML5 bug</a></li>
<li><a href="#item03">3. Media Fragment
Specification</a></li>
<li><a href="#item04">4. Use Cases and
Requirements</a></li>
<li><a href="#item05">5. AOB</a></li>
</ol>
</li>
<li><a href="#ActionSummary">Summary of Action Items</a></li>
</ul>
<hr />
<div class="meeting">
<p class='phone'></p>
<p class='phone'></p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; Date: 24 November
2010</p>
<p class='irc'>&lt;<cite>scribe</cite>&gt; Scribe: Raphael</p>
<p class='irc'>&lt;<cite>scribe</cite>&gt; scribenick:
raphael</p>
<h3 id="item01">1. Admin</h3>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; +1</p>
<p class='phone'>Propose to accept the minutes of last telecon:
<a href=
"http://www.w3.org/2010/11/17-mediafrag-minutes.html">http://www.w3.org/2010/11/17-mediafrag-minutes.html</a></p>
<p class='phone'>+1</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; +1</p>
<p class='phone'>minutes accepted</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; +1</p>
<p class='phone'>ACTION-183?</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; ACTION-183 --
Raphaël Troncy to send reminders to all relevant groups -- due
2010-09-15 -- OPEN</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; <a href=
"http://www.w3.org/2008/WebVideo/Fragments/tracker/actions/183">
http://www.w3.org/2008/WebVideo/Fragments/tracker/actions/183</a></p>
<p class='phone'>close ACTION-183</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; ACTION-183 Send
reminders to all relevant groups closed</p>
<p class='phone'><cite>Raphael:</cite> I did send an email to
whatwg and html5<br />
... but the mail to the html5 mailing list didn't get
through<br />
... subscription problem</p>
<p class='phone'>Silvia, could you please forward my email to
the HTML5 mailing list?</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; oh, did it not go
through?</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; will do</p>
<p class='phone'><cite>scribe:</cite> I have also contacted
Chris Double, Frank Olivier and Eric Carlson to get more
reviews<br />
... and the CSS working group regarding the bug <a href=
"http://www.w3.org/Bugs/Public/show_bug.cgi?id=10723">http://www.w3.org/Bugs/Public/show_bug.cgi?id=10723</a></p>
<p class='phone'>No, silvia, my mail went through the whatwg
mailing list but NOT the html5 one</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; ok</p>
<h3 id="item02">Discuss HTML5 bug</h3>
<p class='phone'><cite>See:</cite> <a href=
"http://www.w3.org/Bugs/Public/show_bug.cgi?id=10723">http://www.w3.org/Bugs/Public/show_bug.cgi?id=10723</a></p>
<p class='phone'>Question is what we should do since this bug
has been closed by Ian</p>
<p class='irc'>&lt;<cite>foolip</cite>&gt; If he's wrong, we
should explain why.</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; I think we need to
propose an actual spec change</p>
<p class='irc'>&lt;<cite>foolip</cite>&gt; That will also do
fine</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; then there is
something to discuss</p>
<p class='phone'><cite>Yves:</cite> for images, if the cropping
is the default rendering behavior, it should be specified in
our spec<br />
... for video and audio, since there is more controls from the
HTML5 spec, if the default behavior is highlighting, then it
could be specified in HTML5</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; (for time
dimension)</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; text about controls:
<a href=
"http://www.whatwg.org/specs/web-apps/current-work/multipage/video.html#attr-media-controls">
http://www.whatwg.org/specs/web-apps/current-work/multipage/video.html#attr-media-controls</a></p>
<p class='phone'><cite>Yves:</cite> silvia, is there in html5,
something that says how external controls are displayed for
video?</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; media-controls should
know how to interact with a fragment</p>
<p class='phone'><cite>Silvia:</cite> there is only
recommendations</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; Yes, I agree - there
should be recommendations on what to display for a
&lt;video&gt; or &lt;audio&gt; element that has a media
fragment URI - in particular since we may recommend a UI
change</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; but we need to
formulate that recommendation</p>
<p class='phone'>I agree Silvia</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; for everything linked
only to the content, the content has to define the behaviour,
for external artefacts, like video controls, as they are part
(even informally) of html5, somehting need to be in the
spec</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; +1 to Silvia</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; and we need to do
this differently for temporal to spatial fragments</p>
<p class='phone'>Problem is also process: bug is closed. Should
we escalate it? Or open a new one for temporal dimension since
we agree on the space dimension</p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; isnt part of ian's
point that there are besides cropping no concrete use cases?
only skimmed very rapidly, so might've misread</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; I'm not 100% sure
about process (maybe Philip knows better), but I think we may
be able to reopen the bug with new information</p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; aryeh's</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; silvia, I think so
too, we need clarification of the intent, and proposed text, as
currently, Ian is perfectly right in closing this bug</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; I agree</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; we haven't provided
any answers to the questions raised in the bug</p>
<p class='phone'>Yes, Thomas is discussing the non-cropping use
case for spatial region</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; the bug was not
registered with a particular focus on spatial fragments - it
was generic for media fragment uris</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; we need to have
changes at least on how the control display should change and
also how the scroll-to-fragid should be done</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; <a href=
"http://dev.w3.org/html5/spec/Overview.html#scroll-to-fragid">http://dev.w3.org/html5/spec/Overview.html#scroll-to-fragid</a></p><a name="action01"
id="action01"></a>
<p class='irc'>&lt;<cite>scribe</cite>&gt;
<strong>ACTION:</strong> silvia to draft the paragraph that the
group will propose to HTML5 regarding how the control of media
fragment URI should be done [recorded in <a href=
"http://www.w3.org/2010/11/24-mediafrag-minutes.html#action01">http://www.w3.org/2010/11/24-mediafrag-minutes.html#action01</a>]</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; Created ACTION-202
- Draft the paragraph that the group will propose to HTML5
regarding how the control of media fragment URI should be done
[on Silvia Pfeiffer - due 2010-12-01].</p>
<h3 id="item03">3. Media Fragment Specification</h3>
<p class='phone'>close ACTION-195</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; ACTION-195 Add a
paragraph in the section 7.1 to specify that video, audio, img
or any href is all treated similarly (range request issued when
facing a media fragment) closed</p>
<p class='phone'><cite>Raphael:</cite> contentious issue, in
the case of a media fragment URI in a very particular context,
i.e. in the audio or video element, AND that media fragment URI
looks like a temporal fragment<br />
... then browser SHOULD?/COULD? issue a range request in a
first place</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; MAY</p>
<p class='phone'><cite>Raphael:</cite> i.e. what we called the
optimistic processing of media fragments<br />
... if the server ack the fact the resource is a video, then it
uses the recipe "Server mapped byte ranges" (section
5.1.2)<br />
... if the server realized that the resource is not a video,
then it ignores the Range header</p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; in this case MAY
sounds too defensive</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; MAY sounds defensive,
but it's the case for all optimisations that are approaching
crossing the layers ;)</p>
<p class='phone'><cite>Raphael:</cite> anytime a URI looks like
a #t= but only if this is the value of the href attibute of
&lt;video&gt;/&lt;audio&gt; or the src attribute of the
&lt;source&gt; element</p>
<p class='phone'><cite>Silvia:</cite> well, Apple implements
the &lt;video&gt; element so that the value could also be a m3u
playlist, not a media element</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; m3u8</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; ok, in that case, time
range won't apply and you will get the whole thing</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; hence the "optimistic
optimization" (and the MAY)</p>
<p class='phone'><cite>Raphael:</cite> I'm curious what will
happen if the video element point to a m3u8 resource in a
browser that is not Safari</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; <a href=
"http://developer.apple.com/library/ios/#technotes/tn2010/tn2224.html">
http://developer.apple.com/library/ios/#technotes/tn2010/tn2224.html</a>
&lt;- example</p>
<p class='phone'>foolip, do you have an opinion on this
discussion?</p>
<p class='phone'>ACTION-191?</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; ACTION-191 -- Yves
Lafon to update the production rules of the time dimension with
the npt format for making the hours optional -- due 2010-11-08
-- OPEN</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; <a href=
"http://www.w3.org/2008/WebVideo/Fragments/tracker/actions/191">
http://www.w3.org/2008/WebVideo/Fragments/tracker/actions/191</a></p>
<p class='phone'>ACTION-173?</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; ACTION-173 -- Yves
Lafon to produce the code that will check the grammar of both
the URI syntax and the Headers syntax -- due 2010-06-22 --
OPEN</p>
<p class='irc'>&lt;<cite>trackbot</cite>&gt; <a href=
"http://www.w3.org/2008/WebVideo/Fragments/tracker/actions/173">
http://www.w3.org/2008/WebVideo/Fragments/tracker/actions/173</a></p>
<h3 id="item04">4. Use Cases and Requirements</h3>
<p class='irc'>&lt;<cite>silvia</cite>&gt; have you collected
all the issues that could be added to the use cases?</p>
<p class='phone'><cite>Raphael:</cite> we have Ericson people
using media fragment URI for video chat, Thomas using the spec
in SemWebVid, HTTP streaming that might use it too<br />
... I wonder if at some point we should not update the UC note
with these initiatives<br />
... silvia, so far I'm only keeping track of the initiatives,
and I think we should report them as well as the issues in the
document later on</p>
<p class='phone'><cite>Thomas:</cite> currently I'm using
simple media fragment to point to temporal sequences of
video<br />
... + content negotiation to get either the RDF annotation of
the sequence (e.g. closed caption) or the video bits<br />
... the overall vision is that search results could include
media fragments URI pointing to sequences that are relevant for
a video search<br />
... use case of finding people and highlighting faces with
media fragments URI, so something more highlighting than
cropping</p>
<p class='phone'><cite>Silvia:</cite> I think you should bring
this to the whatwg mailing list</p>
<p class='phone'><cite>Thomas:</cite> concerned about the too
large traffic of this list</p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; silvia, maybe you
could ping me a link to the thread, and i could jump in</p>
<p class='phone'><cite>Silvia:</cite> I understand, then make
sure to include this in our planned reply to the html5
group<br />
... are you aware of popcorn.js that does similar things that
what you intend to do</p>
<p class='phone'><cite>Thomas:</cite> there is also a couple of
BBC projects that do similar things, twitter streams displayed
in parallel of programs<br />
... also people in DERI working on this<br />
... annotating conference media streams</p>
<p class='phone'><cite>Raphael:</cite> so you agree with the
principle of reporting all these experiments in our UC note at
some point?</p>
<h3 id="item05">5. AOB</h3>
<p class='phone'>none</p>
<p class='irc'>&lt;<cite>silvia</cite>&gt; tomayac, you might
want to read this thread: <a href=
"http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-August/027581.html">
http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-August/027581.html</a></p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; thanks, silvia</p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; can i ask an, erm,
stupid question: if you say html5, do you say whatwg, or
w3c?</p>
<p class='irc'>&lt;<cite>foolip</cite>&gt; tomayac, same same,
whatever comes up first in your favorite search engine :)</p>
<p class='irc'>&lt;<cite>tomayac</cite>&gt; wikipedia ;-)</p>
<p class='phone'>great answer thomas</p>
<p class='phone'><cite>scribe:</cite> the second one should be
W3C</p>
<p class='phone'>:-)</p>
<p class='irc'>&lt;<cite>foolip</cite>&gt; oh, in the context
of &lt;audio&gt;, &lt;video&gt;, I think it's possible to send
a range request without knowing the MIME type, but not really
sane from a purity point of view. In any case I don't see it
happening because the only benefit is one less round-trip, and
only works with specialized servers</p>
<p class='irc'>&lt;<cite>foolip</cite>&gt; I don't think the
benefit is tangible enough that browsers or servers would
bother implementing it.</p>
<p class='irc'>&lt;<cite>Yves</cite>&gt; it will depend on big
content provider interested or not in that (to reduce the
load)</p>
<p class='irc'>&lt;<cite>foolip</cite>&gt; right, should any
browser or provider show any kind of interest in it I could of
course reconsider</p>
<p class='phone'>you represent ONE browser, and I'm talking to
provider, such as Dailymotion, who could be interested ... so
it might be worth considering at some point</p>
</div>
<h2><a name="ActionSummary" id="ActionSummary">Summary of Action
Items</a></h2><!-- Action Items -->
<strong>[DONE]</strong> <strong>ACTION:</strong> silvia to draft
the paragraph that the group will propose to HTML5 regarding how
the control of media fragment URI should be [recorded in <a href=
"http://www.w3.org/2010/11/24-mediafrag-minutes.html#action01">http://www.w3.org/2010/11/24-mediafrag-minutes.html#action01</a>]<br />
&nbsp;<br />
[End of minutes]<br />
<hr />
<address>
Minutes formatted by David Booth's <a href=
"http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm">
scribe.perl</a> version 1.135 (<a href=
"http://dev.w3.org/cvsweb/2002/scribe/">CVS log</a>)<br />
$Date: 2010/11/24 11:28:45 $
</address>
<div class="diagnostics"></div>
</body>
</html>