See also: IRC log
<trackbot> Date: 10 December 2008
<raphael> Raphael: we have discussed mainly 3 topics
<raphael> ... Existing technologies: survey of the various syntaxes
<raphael> ... Different dimension for addressing fragments: name, time, space, track
<raphael> ... explain further what we mean by tracks, i.e. what is available in the container format
<raphael> ... discuss implementation issues dealing with caching
<raphael> See: http://www.w3.org/2008/WebVideo/Fragments/wiki/Syntax
<davy> scribenick: davy
Raphael at the drawing board
four different dimensions: time, space, track, name
raphael: do we allow any combination of these dimensions
jack: shouldn't we use 'chapter' instead of 'name'?
raphael: which combinations do we allow?
jack: use either name or a
combination of time, space, and track
... multiple video tracks in combination with space will cause problems
... because there is no single resolution within the media resource (in contrast to the time axis)
raphael: do first the time or
... afterwards, do the spatial clipping
... we do not have any commutative problems because we specify the processing order of the axex
jack: ambiguous cases are multiple track selections in combination with spatial clipping
How to specify the values of the different dimensions?
<erik> scribenick: erik
* time representation
<tom> "The sub-frame is sometimes referred to as a field in the context of synchronization with an interlaced video media object."
scribe: npt and smpte are essential
<scribe> ACTION: jack to describe why smpte is the only frame accurate one [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action01]
<trackbot> Created ACTION-25 - Describe why smpte is the only frame accurate one [on Jack Jansen - due 2008-12-17].
jack: caching server will not only want to record byte ranges, but also time info
(and so thinks Yves)
raphael: assumption, "&" will be our first separator (and try to avoid "/")
<davy> yves: if you specify a fragment until the end of the stream, you specify a start, followed by a separator, followed by nothing
scribe: begin time = download all
... begin time "separator" = download from begin time to the end and play
... begin time "separator" [end time] = download from begin time to end time and play
... [begin time] "separator" end time = download from beginning to end time and play
second separator: ","
<Gui> hi all
<Gui> Let me familiarize myself with the agenda. I will only be able to attend for a short while, I am expected at a team meeting this afternoon
* name representation
jack: if naming is available in container formats, what can be used?
<Yves> anything could be used provided it's % escaped in the URI
jack: http://www.ietf.org/rfc/rfc3986.txt ... 2.3 unreserved characters
<nessy> me btw "/" has no reserved purpose inside query or fragment as per that rfc
<davy> two solutions for internationalization issue for named fragments
<davy> ACTION: Davy to look at various container formats what kind of characters they allow to name their tracks/fragments [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action02]
<trackbot> Created ACTION-26 - Look at various container formats what kind of characters they allow to name their tracks/fragments [on Davy Van Deursen - due 2008-12-17].
<davy> 1. quotes + characters + % escaped
<davy> 2. only use the unreserved characters (specified in URI 3986, sect. 2.3)
<davy> we need to specify the encoding used to define the name
<davy> raphael: we will ask Felix about this internationalization issue
* track representation
scribe: track="some alphanumeric
... what's between quotes depends on container format
Jack: do we want to predefine track names? ... very easy e.g. track="audio"
<Gui> that could be left to the user agent to determine
<Silvia> sakim, mute me
Jack: predefined track names could be: audio, video, subtitle
yves: do tracks for instance always have id's?
resolution: we only allow tracknames that the underlying container formats know
jack: is it mandatory to have a kind of ROE-file for finding out the capabilities of a resource?
<Gui> the user agent may have schemes that will allow it to associate tracks ID to generic names such as Audio for each media type. Can track IDs or structuring help determine their type?
<Gui> Examples : Audio track track=AU184, track=AU185 become track=audio1 or track=audio2 for easy addressing, the user agent will do the translation
<Gui> Yes please!
<Gui> Silvia: It is difficult to follow on the phone, summary please
<Gui> Silvia - I am trying to follow from South Africa
<erik2> * space representation
<erik2> brainstorming ...
<erik2> ... resolution: last F2F ... only rectangle for v1
<erik2> jack: not using the name "rect", please ... cause ambiguous ... rather have "xywh" instead
<Gui> Silvia agree with Jack xywh or TRBL ?
<erik2> Jack's proposal: xywh=20,20,40,40
<Gui> trbl=20,20,40,40 means the margins are 20,20,40,40
<Gui> xywh=20,20,40,40 means the pixel position 20,20 then offset 40 vertical and horizontal
<Gui> Jack about assuming coordinate systems ( (0,0) at the top) and units (pixels or inches?)
<erik2> extra questions: what's the coordinate system and/or metrics?
<erik2> proposal syntax: [xywh=int,int,int,int]
<erik2> proposal coordinate system: [0,0]=top-left
<Gui> Typically this is a rectangular crop operation. I'm not sure that a Crop operation is best described using xywh, TRBL is clearer when cropping
<erik2> proposal units: default=pixels
<Silvia> with xywh we also need to specify whether w goes left/right and h goes up/down for xy position
<erik2> ... within v1 no alternatives allowed like cm,inch,point,%
<Gui> with xywh you have a problem that you could ask to crop regions that are out of range. With TRBL one cannot
<Gui> T margin-top, R margin-right, B margin-bottom, L margin-left,
<Gui> but again, the ordering may be a problem.
<Gui> we can choose a scheme and then user agents can do the conversion ... between cropping schemes
<Silvia> I thought trbl meant (top,right) as start point and (bottom,left) as end point, offset from [0,0]
<erik2> suggestion: TRBL
<erik2> ... pro: not possible to crop outside media
<Gui> very confusing, yes Silvia
<erik2> ... con: not intuitive
<erik2> suggestion: x1y1x2y2
<Silvia> I like x1y1x2y2, but I don't like this name :)
<Gui> What's nice about the name is that it gives the ordering clearly
<Gui> another good feature : you don't always have to specify all the coordinates or all the margin crops
<erik2> jack: that is just not a good feature
<Gui> Jack : 4 separate parameters rather than one compound one
<Gui> Silvia: Name suggestion for x1y1x2y2 ?
<Gui> Well the other name was TopLeft-BottomRight
<Silvia> cartesian ?
<Silvia> cartesian coordinates?
<erik2> final suggestions:
<Silvia> x1y1x2y2=int,int,int,int [pixels]
<erik2> or xy1xy2
<Gui> In practice: Practically with video image cropping, you want to make sure that the width and height AND the positions are multiples (of 4 for example).
<erik2> or xyxy
<Gui> Jack: I remember you talking about it!
<Gui> no objection to x1y1x2y2
<erik2> resolution: go for x1y1x2y2 !
<jackjansen> Time currently has: [format:]begin, end
<Silvia> is that time=[format:]begin,end ?
<jackjansen> So space could have space=[unit:]x1, y1, x2, y2
<jackjansen> Then default for format would be "pixel" but allow for extension later
<Silvia> possible confusion about format -> LTRB/x1y1x2y2 and format -> unit
<Gui> Jack, are you talking about relative dimensions such as % and em ?
<erik2> raphael: good idea to NOT mix units
<Gui> What are the units proposed?
<erik2> ... and they should all be possitive integers
<Yves> gui, yes, we are talking also about %
<Yves> (and parsecs for floating number examples)
<erik2> ACTION: yves to propose by email a good list of units that can be used [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action03]
<trackbot> Created ACTION-27 - Propose by email a good list of units that can be used [on Yves Lafon - due 2008-12-17].
<Gui> Fragments 11 Annotations 0 http://www.googlefight.com/index.php?lang=en_GB&word1=%22links+to+fragments%22&word2=%22links+to+annotations%22
<Gui> Bon Appetit
<Gui> Attending a meeting, brb.
jack: http://www.w3.org/2008/WebVideo/Fragments/wiki/Syntax (thx Raphael for quick summary)
<raphael> scribenick raphael
<raphael> We are going through the summary
<raphael> ACTION: Jack to suggest an exhaustive list of possible values for (temporal) format in the proposed syntax [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action04]
<trackbot> Created ACTION-28 - Suggest an exhaustive list of possible values for (temporal) format in the proposed syntax [on Jack Jansen - due 2008-12-17].
<raphael> Discussion on separators: primary can be '&' or ';' secondary can be ','
<raphael> Jack: I suggest we choose one in the WD but put in bold the question, which one we should use
<raphael> Yves: I would almost prefer ';' instead of '&'
<raphael> ACTION: Erik and Raphael to organize an informal poll what is the best separator [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action05]
<trackbot> Created ACTION-29 - And Raphael to organize an informal poll what is the best separator [on Erik Mannens - due 2008-12-17].
<raphael> Yves: should the value of the name dimension quoted?
<raphael> ... i would say: http://www.example.com/movie.mov#name="the%20kiss%20scene"
<raphael> Jack: I hate x1y1x2y2!
<raphael> ... I would almost prefer xywh
<raphael> ACTION: Erik and Raphael organize an informal poll what is the best way of specifying rectangle [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action06]
<trackbot> Created ACTION-30 - And Raphael organize an informal poll what is the best way of specifying rectangle [on Erik Mannens - due 2008-12-17].
<raphael> Yves: RFC 3986 says that simple quote is a sub-delimiter in a URI
<raphael> ... but double quotes should be escaped
<Yves> (allowed in fragments: ALPHA / DIGIT / "-" / "." / "_" / "~" / "!" / "$" / "&" / "'" / "(" / ")" / "*" / "+" / "," / ";" / "="
<Yves> everything else has to be %-escapes
<raphael> Jack: do we want "time=" or "t="?
<raphael> All: we tend to prefer "t="
<raphael> Davy: use "id=" instead = "name=" ... as CMML
<raphael> ALL: we tend to prefer "id="
<raphael> Yves: we should also quote track value, as they can come from an unbounded set of caracters
<raphael> All: agree
<raphael> scribenick: raphael
Raphael: we welcome the assistance from Felix
Felix: you should use UTF-8, and %-escape caracters as per RFC3987
Jack: I thought that %-escape were latin-1 caracters
Felix: a procedure is described for translating any unicode caracters into %-escape UTF-8 characters
Jack: in this case, we can just assume that the identifiers are UTF-8 encoded
Raphael: so the syntax does not need to include a particular encoding
Jack: if a movie has chapter names in Japanese, then we can have a UTF-8 translation of this name, either using a %-escape or not
<jackjansen> am I still in?
Raphael: the answer is then provided by either URIs, or IRIs RFC (check later LEIRIs)
<trackbot> ACTION-14 Review the use cases and requirements document closed
<jackjansen> am I alive?
<scribe> ACTION: Erik and Raphael to organize a poll to see whether the quotes around track and id are optional [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action07]
<trackbot> Created ACTION-31 - And Raphael to organize a poll to see whether the quotes around track and id are optional [on Erik Mannens - due 2008-12-17].
Jack: I think quotes should be optional
Yves: I think quotes should be mandatory
Raphael: future telecon: 17/12
... no telecon on 24/12 and 31/12
<jackjansen> xmlspec: http://www.w3.org/XML/1998/06/xmlspec-report-v21.htm
<scribe> ACTION: Yves to check and grant cvs access to the key members of the group (erik, raphael, jack, silvia) [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action08]
<trackbot> Created ACTION-32 - Check and grant cvs access to the key members of the group (erik, raphael, jack, silvia) [on Yves Lafon - due 2008-12-17].
<erik> Topic 6: first document
<erik> 1. intro
<erik> 2. Use Cases
<erik> 3. Requirements (see page + types of fragment addressing)
<erik> 4. Naming Fragment (identification)
<erik> ... # vs ?
<erik> ... details of formal grammar & semantics
<erik> ... TODO interaction with other standards (SVG, SMIL, ...) and extreme cases
<erik> 5. Retrieving Fragment
<erik> ... HTTP 2-way & 4-way handshake
<erik> ... caches/proxies
<erik> ... client-side requirements
<erik> 3.5 Technologies Survey
<erik> who will write initial chapters?
<erik> ... 5. Raphael
<erik> ... 4. Jack
<erik> ... 3.5 Erik
<erik> ... 3. Silvia/Raphael
<erik> ... 2. Silvia
<erik> ... 1. Intro ???
<erik> ... starting up Raphael/Erik
<scribe> ACTION: Erik and Raphael to start the template of the document [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action09]
<trackbot> Created ACTION-33 - And Raphael to start the template of the document [on Erik Mannens - due 2008-12-17].
Raphael: let's talk about implementation
Yves: i can work on
implementation from February
... not before
Davy: we have a server pretty independant of encoding format, we can work together
Jack: on client side, the easiest thing to do will work on a smil player, and just deal with a temporal selection
Raphael: will, for example, a firefox plugin be able to generate good http packets and communicate with the server ?
Jack: why we don't use a
python-url library, or curl ?
... then we are independant of the browser and replicate its behaviour
Yves: I can talk to the curl
maintainer on the html mailing list, to see whether curl is
doing some syntax checking
... otherwise, we can do that with a wget
Jack: so our first implementation will be a shell script on client side
<scribe> ACTION: Jack to look at python-url library to see whether he could implement the logic on client side [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action10]
<trackbot> Created ACTION-34 - Look at python-url library to see whether he could implement the logic on client side [on Jack Jansen - due 2008-12-17].
<scribe> ACTION: Raphael to look at curl and/or wget to see whether the logic could be implemented on client side [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action11]
<trackbot> Sorry, couldn't find user - Raphael
<jackjansen> tracknot, who is here?
<jackjansen> trackbot, who is here?
<trackbot> Sorry, jackjansen, I don't understand 'trackbot, who is here?'. Please refer to http://www.w3.org/2005/06/tracker/irc for help
<Yves> trackbot, status
<jackjansen> ACTION: Raphaël to look at curl and/or wget to see whether the logic could be implemented on client side [recorded in http://www.w3.org/2008/12/10-mediafrag-minutes.html#action12]
<trackbot> Created ACTION-35 - Look at curl and/or wget to see whether the logic could be implemented on client side [on Raphaël Troncy - due 2008-12-17].
<Yves> ACTION -11
<trackbot> Sorry, bad ACTION syntax
The room thanks IBBT for the brilliant organization
This is scribe.perl Revision: 1.133 of Date: 2008/01/18 18:48:51 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/...discuss/... discuss/ Succeeded: s/ccaching/caching/ Succeeded: s/and name/and track/ Succeeded: s/download only the sequence and play/download from begin time to the end and play/ Succeeded: s/begin time "separator" end time/begin time "separator" [end time]/ Succeeded: s/text/subtitle/ Succeeded: s/xyzh/xywh/ Succeeded: s/xwwh/xywh/ Succeeded: s/ of 4 (for example)/ (of 4 for example)/ Succeeded: s/space could have [format:]/space could have [unit:]/ Succeeded: s/[unit:]/space=[unit:]/ Succeeded: s/ted/tend/ Succeeded: s/RFC3986/RFC3987/ Succeeded: s/caracters/characters/ Succeeded: s/...)/...) and extreme cases/ Succeeded: s/4./... 4./ Succeeded: s/3. 5/... 3.5/ Found ScribeNick: davy Found ScribeNick: erik Found ScribeNick: raphael Inferring Scribes: davy, erik, raphael Scribes: davy, erik, raphael ScribeNicks: davy, erik, raphael Default Present: +2712841aaaa, Guillaume, +329331aabb, Frank, Davy, Erik, Raphael, Tom, Wim, Jack, Yves, +61.2.801.2.aacc, Silvia Present: Jack Yves Erik Frank_(canon_observer) Raphael Davy Tom Wim Silvia_(remote) Regrets: Michael Agenda: http://www.w3.org/2008/WebVideo/Fragments/wiki/SecondF2FAgenda Found Date: 10 Dec 2008 Guessing minutes URL: http://www.w3.org/2008/12/10-mediafrag-minutes.html People with action items: davy erik jack l rapha raphael yves[End of scribe.perl diagnostic output]