401
edits
Changes
added Henri's comments
* Another way would be to provide a javascript API for the text elements.
DISCUSSION:
Exposing timed text to scripts as DOM nodes is an interesting idea.
However, exposing them as direct subelements is probably not the way to go for two reasons:
1) The general expectation is that the DOM tree is the parse tree of one HTTP resource--that is, there's one Document object per HTTP resource.
2) For security reasons, allowing a video file cause node insertion into an HTML DOM tree seems like a potentially dangerous thing.
If a DOM tree representation is natural for the captioning format, it should probably be available under a separate Document object like iframe content.
OTOH, the DOM is rather heavyweight, so if the caption renderer isn't DOM-based, it would probably be better to expose a lighter specialized JS API.
=== Specification needs ===
