Re: <track> and WebVTT in Opera Next

2012/8/8 Philip Jägenstedt<philipj@opera.com <mailto:philipj@opera.com>>

        I found a couple of (maybe known) issues related to WebVTT
        classes and
        settings.
        Taking this video as an
        example:http://www.delphiki.com/html5/playr/
        <http://www.delphiki.com/html5/playr/>
        The first cue (see webvtt file:
        http://www.delphiki.com/html5/playr/examples/dw_trailer_low.vtt
        <http://www.delphiki.com/html5/playr/examples/dw_trailer_low.vtt>)
        does not
        render properly, multiples classes seems to be the problem.


    I investigated a bit and it turns out that the problem is
    position:0%. Per spec, that means that the maximum size of the cue
    should be 0, which is really strange. I filed a
    bug:https://www.w3.org/Bugs/Public/show_bug.cgi?id=18501
    <https://www.w3.org/Bugs/Public/show_bug.cgi?id=18501>

    If you try position:0% line:0 you'll see where the rest of the text
    ended up -- one word per line! The reason that the cue wasn't moved
    up into the viewport is because the cue box has width 0 and it's not
    really clear if that should be considered intersecting the viewport
    or not :/


In this case, I may have misinterpreted the spec about position.
I thought this was the way it should behave: 
http://leanbackplayer.com/other/webvtt.html#cue-settings-textposition
I think I get the general idea now.
Anyway this isn't the clearest part of the spec.

        ::cue:past & ::cue:future pseudo-classes don't seem to be supported.


    Those have been implemented and tested, can you give an example of
    something that doesn't work?


I haven't been able to make them work.
I may not have done it the right way, see the last 2 selectors in this 
file: http://www.delphiki.com/html5/playr/style.css

        Could you list the features implemented?


    We've tried to implement everything except vertical text and ruby.
    There are also some deviations from the spec where we've filed spec
    bugs, and the pauseOnExit flag isn't frame-accurate.


Thanks for the heads up!

        Non-related: I'm not sure I mentionned it on the mailing list,
        but I made
        this little srt to vtt converter a while back:
        http://atelier.u-sub.net/srt2vtt/
        <http://atelier.u-sub.net/srt2vtt/>


    I tried feeding it a file encoded as Latin-1 to see if it would
    guess or ask, but was told that it was not a proper SubRip file.
    Figuring out the correct encoding is an important part of
    conversion, so it would be nice if in this case the user was asked
    to preview likely encodings and pick the correct one.


Actually, the script is supposed to detect the encoding, I'll have a 
look at your specific case.
Source is available here: https://github.com/delphiki/SubRip-File-Parser