[Bug 26925] [InbandTracks] DASH Text Track Mapping

https://www.w3.org/Bugs/Public/show_bug.cgi?id=26925

Bob Lund <b.lund@cablelabs.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |b.lund@cablelabs.com

--- Comment #1 from Bob Lund <b.lund@cablelabs.com> ---
(In reply to Cyril Concolato from comment #0)
> The current text in "5. Mapping text track content into text track cues" is
> completely ununderstandable. 
> 
> What does the following sentence mean?
> "DASH text track data may be contained in files or encapsulated in MPEG
> media container formats ISOBMFF and MPEG-2 TS."
> There is no such thing as a "DASH text track". What does it mean that "data
> may be contained in files"? What files are we talking about here: MPD? What
> data are we talking about: media data?
> What is the purpose of the first paragraph ? It has no normative statement.
> What should UA or applications do with it?
> Suggestion: remove that paragraph.

Sentence is important because it defines the scope of media formats in DASH
from which text track cues are sourced. The sentence has been simplified to
improve readability.

> 
> That other sentence does not make sense:
> "DASH text track data is in a file in the WebVTT format if the file MIME
> type is "text/vtt" and should be exposed as a VTTCue as defined in [WEBVTT]"
> The first part is a tautology: data in a file is of WebVTT format if the
> MIME is text/vtt. That's by definition of WebVTT. Does the sentence use the
> "if" in the wrong place ? Do you want to say the following:
> "If a DASH MPD references content identified by the MIME type "text/vtt",
> that content shall be exposed as a TextTrack using VTTCue as defined in
> [WEBVTT]" 
> The same is true for the next sentence for TTML.

The semantics of your suggestion and the existing text are identical. The
sentences can be simplified, e.g Media content with the MIME type "text/vtt" is
in the WebVTT format should be exposed as a VTTCue as defined in [WEBVTT]. It's
to identify the particular text track format, rather than just citing a
reference, because some text track formats don't define a Cue API.

> 
> Why are those statements about WebVTT or TTML in the DASH section anyway? 

Because the section is about creating TextTrackCue from DASH content and WebVTT
and TTML are formats that are relevant.

> It should be applicable to any container/manifest format (e.g HLS, ...). 
> Suggestion: Remove WebVTT and TTML specific content from the DASH section
> and create separate sections.
> 
> The sentence about MP4 has the same problem plus others:
> " "application/mp4" (ISOBMFF container with one track) " is wrong. MP4 files
> advertised with a MIME "application/mp4" may have multiple tracks. RFC 4337
> indicates (in guidelines not in normative statement) that it should not
> contain audio or visual material. It may contain multiple MPEG-7 tracks or
> else.

Good point.

> Suggestion: Rephrase saying:
> "If content is identified by a MIME type used in section ISOBMFF, the
> mapping in that section applies".
> 
> Same problem with the sentence about MPEG-2 TS. Same suggestion.

Good point.
> 
> I can make all those changes in a pull request if you want.

PR [1] that makes changes that address your comments.

[1] http://rawgit.com/boblund/HTMLSourcingInbandTracks/bug26925/index.html

-- 
You are receiving this mail because:
You are the QA Contact for the bug.

Received on Thursday, 2 October 2014 15:39:29 UTC