RE: TT Content Buffering and Timing Scenarios

See inline.

> -----Original Message-----
> From: Charles Wiltgen [mailto:lists@wiltgen.net] 
> Sent: Thursday, January 30, 2003 2:03 PM
> To: List EUR W3C Timed Text
> Subject: Re: TT Content Buffering and Timing Scenarios
> 
> Glenn A. Adams wrote...
> 
> >> (2) Live (created-on-the-fly) timed text can never have 
> XML rules enforced
> >> because the entire file doesn't even exist when it starts 
> to be shown.  Live
> >> RealText provided me with a lot of challenges in other 
> ways as well.
> > 
> > Could you describe some of the issues with "live timed 
> text" more fully and
> > how it can "never have XML rules enforced"? Are you saying that even
> > well-formedness rules aren't possible?
> 
> Not by the client when it's streamed, since you have no idea 
> what the rest
> of the file looks like.  The server can enforce 
> well-formedness, though,
> and anybody using TT for the wire format as well can either send TT
> before streaming content (this is similar to a common technique for
> QuickTime) or stream well-formed TT chunks in realtime as necessary.

Actually, I can imagine a number of ways to handle this and still
satisfy XML well-formedness rules. Here are some ideas:

1. Each "access unit" could be a complete XML document. In this
case, WFC and VC could be checked on each access unit.

2. An unbounded sequence of access units could be construed as an
external entity, where each access unit constitutes an entire
top-level element. [External entities can have multiple top-level
elements.] An XML document template that implicitly references
that external entity could be implied in the client, e.g., (making
up some temporary vocabulary):

<tt:stream-wrapper xmlns:tt="...">
&stream;
</tt:stream-wrapper>

where one would effectively have:

<!ENTITY stream "
<tt:access-unit>...</tt:access-unit>
<tt:access-unit>...</tt:access-unit>
<tt:access-unit>...</tt:access-unit>
...
">

In this case, WFC and VC are, by definition, satisfied on the
implied wrapper, and WFC and VC could be checked on each
access-unit independently over time as they arrive.

Comments?

G.

Received on Thursday, 30 January 2003 14:34:22 UTC