[
Lists Home |
Date Index |
Thread Index
]
> > As I noted, the speedup is most likely because the cost of building
> > a DOM is usually in object construction, rather than in parsing.
>
> Ahh, I see. Yes, I take it back, Patrick's observations *are* an
> effective demonstration that building smaller DOMs takes substantially
> less time than building larger DOMs from the same source document and
> therefore that pre-processing to remove (or turn into text nodes) data
> that you're not interested in is a good tactic.
I am wondering why there is no API combining Push or Pull with DOM.
Most DOM parsers have an underlying event parser anyway.
That way you can hook yourself into the building of the DOM tree
and choose which nodes to keep, and which ones to discard or transform.
Might be an idea for the new XOM API.
Karl
|