Gregor J. Rothfuss wrote: ... > it turns out though that i usually have to run forrest
multiple times until everything is just like i want it, and with changes.html already built (and not having changed, nor the navigation having changed) there is really no need to rebuild it.
Ok, I think I understand.
Cocoon CLI AFAIK does not yet check if the file is to be regenerated or not. It's not so simple as it seems, as the links are crawled, so it would need to cache the links in every page to a static location to be able to reuse them later when not regenerating the page.
I agree that it's needed.
from what i can tell, forrest spends a lot of time pondering the thousands of external links on that page (to SVN diffs)
Are the svn diffs local? It should not go and look for public links, if it does it's wrong.
you can view the result at
http://lenya.apache.org/changes.html
the links are not local.
What puzzles me is that it should not ponder anything on these links. Darn, what is it doing? :-/
Have you tried using a newer xalan and xerces version?
haven't yet. i doubt it's a transform issue. if i take out the generation of the diff links for the initial status.xml
from
http://svn.apache.org/viewcvs.cgi/*checkout*/lenya/docu/tools/svn2changes.xsl?rev=154862
<link href="http://svn.apache.org/[EMAIL PROTECTED]&view=rev">Diff</link>
forrest is *much* faster, which leads me to suspect it is trying to lookup something.
Very strange indeed.
Does the file take so long also doing a traqnsform ouside of Forrest?
the initial svn 2 changes is quite fast, but then again its not a secret that saxon is way faster than xalan.
The question was because I read that Axis was having problems with a JDK version of Xalan that was taking an order of magnitude more time to do something. Maybe I read wrong, I don't remember the page.
I hope I have some time tomorrow to try it and debug it myself.
--
Nicola Ken Barozzi [EMAIL PROTECTED]
- verba volant, scripta manent -
(discussions get forgotten, just code remains)
---------------------------------------------------------------------