[
Lists Home |
Date Index |
Thread Index
]
* Karl Waclawek <karl@waclawek.net> [2005-01-17 09:20]:
> Alan Gutierrez wrote:
>
> >
> > Hmm... Okay, sorry for the verbose example, it's what I'm
> > working on right now.
> >
> > The idea is that in order to place the exploaded element, I need
> > to select:
> >
> > /request/path-info/path/text()
> >
> > And I need to match:
> >
> > /request/path-info[path]
> >
> > Or maybe even:
> >
> > path[contains(text(), '/')]
> >
> > Which strikes me as resonable, but not, to my understanding, a
> > question of look *ahead*. More like look behind.
>
> Why not just match on /request/path-info/path and process the
> accumulated character data in the end-tag event for this element?
> Isn't that simple enough?
Sure. I think you and I are talking about matching a pattern
as a series of events.
The last time I saw this discussion go by, it seemed like people
were thinking of the problem in terms of document nodes, and
felt that only the ancestor axis, and maybe the preceeding
sibling axis of each ancestor were avaiable for traversal.
http://lists.xml.org/archives/xml-dev/200412/msg00672.html
"Essentially in order to make a useful case of this (surrendering
at baz3) you would need to cache all previous events so that
they could be replayed when the predicate's boolean value
was determined. Unfortunately that solution doesn't win
much against the DOM or other tree based models."
I'm coming back with the notion that a dfa would win against DOM
in this case, but I hadn't time to do the homework in December.
--
Alan Gutierrez - alan@engrm.com
|
