See also: IRC log
<gmandyam> Giri Mandyam, Qualcomm, joining HTMLCue breakout session
cribe: giuseppe
<scribe> scribe: giuseppe
Nigel: in the TTWG we are looking at two formats
TTML and WebVTT
for webvtt we have a cue, with in/out and style
when we look at other formats
we are asking ourselves how we implement them in the rbrowser
dash.js is trying to use texttrackcue to support other tt formats
but to do that you need to overload the texttrackcue interface
because that is not designed to handle something else then webvtt
the question is why do we need a specific vtt cue, can we have a more generic cue mechanism that can support more formats
for example can we have an html cue
that only have in/out and an html payload
that the browser can attach that html payload on in/out
or better attach on in, remove on out
the idea with texttrack is to separate matadata in different tracks
in ttml structure mixes different things in one format (text and style and metadata)
status of this is that htmlcue is an idea
ttwg presented the idea to the whatwg
the feedback was there could be some unintended consequences
example what happens if the html contains a script
or what happens to the browser context
so I'm interested to discuss these issues and how to appraoch the problem
giuseppe: is the main goal here to support ttml?
nigel: yes mainly, but not only
ttml could be one source of information for the htmlcue track
simon: one issue with this
is security, if you allow all html in the htmlcue
subtitles usually don't have scripting
the other issue is if we assume thatw e need to subset html, does it still support the use cases you want to address
currently with wbvtt you can use a metadata track
and you can use javascript and render it
and there you don't have the security problem
why can't you use the webvtt metadata track and expose ttml then render in the page?
giri: the problem is, the html
spec allows for 250ms delay
... and this might be an issue for things like subtitles
giuseppe: does it mean we need a high priority cue?
simon: are you sure this delay of 250ms apply to the cues?
(room checking the spec)
nigel: so the requirement is that accuracy has to be greater then 250ms
simon: the 250ms is only for the timesupdate
but should not apply to the cues
jc: cyril has done a demo using the events
x1: the secureity issue here i major
maybe is better to look at extending the current api
giri: what about the idea of using a sandbox
nigel: you could block any script and maybe threat it like an iframe
simon: the metadata kind is designed to support this use case
xxx: could there be a problem of performance, is the html is too complex?
jc: maybe you can do some testing and show if there is really a performance problem, and if you really need some htmlcue
nigel: so the proposed model is to use existing track cue with metadata kind, although is not metadata, and then render in javascript
xxx: there are of course some problems
for example the browse might need to expose the subtitles preferences
so that the js renderer know what to do
the other problem is that you cannot use the defaul player controller
but if you are doing this custom subtitlles handling you can probably build your custom controls as well
giri: what happens if kind is set to something other then the allowed values
simon: I think it defaul to subtitles
andreas: the question is still, wouldn't be better to have some easier way to support this
also, it it's easier for people to transalte from ttml to html then to vtt
because for webvtt you need to know both the sintax but also the renderin model that is different
all: discussion about if it's
feasible to make a subset of html suitable for this
htmlcue
... some people concerned that subsetting html has been done
before and always failed
we should not try to do it again
nigel: isn't this very similar to sandboxing?
simon: one of the reason I think this si different, is because there is no way to disable fetching external resources, and I think we need that
nigel: why should we do that, images might be needed
simon: but then you might have a privacy issue
e.g. an external resource/server could now monitor videoplayout events, monitor what you watch
something that was not possible before (only the internal scripting of the page could do it)
xxx: the image issue could be solved using data: urls
<scribe> ACTION: file a feature request to expose user preferences about subtitles to the page [recorded in http://www.w3.org/2015/10/28-htmlcue-minutes.html#action01]
<scribe> ACTION: work on a prototype using the metadata kind for texttrack and to try and highlight limitation and issues [recorded in http://www.w3.org/2015/10/28-htmlcue-minutes.html#action02]
This is scribe.perl Revision: 1.140 of Date: 2014-11-06 18:16:30 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/preent/present/ Succeeded: s/xxx/andreas/ Succeeded: s/resource/issue/ Succeeded: s/ruls/urls/ Succeeded: s/fiel/file/ Found Scribe: giuseppe Inferring ScribeNick: giuseppe WARNING: No "Topic:" lines found. Present: nigel hober Jean-Claude_Dufourd giuseppe JP Abello Giri Mandyam Kurosawa Taketoshi Yamane Kinji Matsumura Akitsugu Baba glenn_adams jerry_smith Got date from IRC log name: 28 Oct 2015 Guessing minutes URL: http://www.w3.org/2015/10/28-htmlcue-minutes.html People with action items: file work 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]