cocoon-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Derbyshire <derbysh...@globalserve.net>
Subject Re: Serious error in Xalan -- parsing stylesheet PIs -- URGENT.
Date Sat, 27 May 2000 02:35:23 GMT
At 09:38 AM 5/26/00 +0200, you wrote:
>Paul Derbyshire wrote:
>> 
>> >You might want to read up on that. A leading / is exactly within the
>> >standard and means exactly the document root.
>> 
>> But it's not truly relative, as I explained. If I moved to another server
>> and the path to my user directory changed from /users/jrandom to /~jruser
>> I'd have to change every internal link in my site, and if the site is big
>> enough that's impractical to expect of me.
>
>No

Wrong.

If I move from a site where my directory on the Web server is
/users/jrandom to one where it is
/~jruser and leave a link that formerly worked as <a
href=/users/jrandom/foo.html>, that link will
start to be a 404, because the address http://new.isp.com/users/jrandom
doesn't exist, unlike http://old.isp.com.
The only way this kind of URL would never have to be changed would be
moving from a server where my site is rooted at the server document root to
another server where my site gets rooted at the server document root, and
the only time you can ever have your site at the server's document root is
when it's *your server*. I don't have one. Yet.

What could work would be to use a PURL for every link, even local ones; but
this has its own problems, such as the unneeded extra load put on the PURL
resolver and the dependence on a PURL resolver (which could prove to be
less than 100% reliable, and which would demand that the site file
hierarchy be uploaded to the server just to test it. I want to be able to
test the site locally without even having to make an Internet connection,
using just the batch-XML-to-HTML program and a Web browser.

>Why not write a script that runs cocoon repeatedly?

You're joking, right? Can you imagine how slow that would be, a full-blown
JVM startup for every single new or updated file? The JVM takes ages to
bootstrap, which is why I'm using a Java app to call a Xalan API, thus
meaning that there is only one JVM startup for every *batch* of possibly
hundreds of updated and new files.

>In any case, what I would do is install Apache and JServ and grab the HTML
pages automatically with wget.

An ugly hack that requires learning a bunch of new pieces of software
before I'm particularly ready, and requires going to Linux for every test
and laboriously rebooting to Windows for every network access -- either to
upload the new and updated files to a server or to use email or surf the
Web or download things.

One step at a time. Get the XML to HTML working; get a starting library of
style sheets working; get some content up and tested; and then worry about
running my own Web server, figuring out how to do servlets, getting that
broadband connection, etc.


-- 
   .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
-()  <  circles, and bark is not smooth, nor does lightning travel in a
   `*'  straight line."    -------------------------------------------------
        -- B. Mandelbrot  |http://surf.to/pgd.net derbyshire@globalserve.net
_____________________ ____|________                          Paul Derbyshire
Programmer & Humanist|ICQ: 10423848|

Mime
View raw message