cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tagunov Anthony" <>
Subject Re: the "evil" document() function
Date Sun, 18 Mar 2001 15:33:23 GMT
On Sat, 17 Mar 2001 16:16:42 +0100, Peter C. Verhage wrote:

>----- Original Message -----
>From: "Tagunov Anthony" <>
>To: <>
>Sent: Friday, March 16, 2001 10:11 PM
>Subject: the "evil" document() function
>: On Fri, 16 Mar 2001 14:56:07 -0500 (EST), Donald Ball wrote:
>: >... the document() function as provided by xalan
>: >caches.... aggressively, let's say. forever is another word you could
>: Donald, maybe the use of this function should be generally discoureged?
>Why? Just because it does not work entirely as people expect it to do? I
>think the document() function can be really usefull for a lot of things at
>which you rather don't use Xinclude, but it would be indeed nice if this
>function would check if there it's a new version or not... How difficult is
>this issue, because the XInlcude processor can do it, so why can't the folks
>at Xalan think of a similar approach for the document() function?

Unfortunantly looks quite difficult: to make the Cocoon caching system work
okay the XSLT processor should somehow let _Cocoon_ know what
files it has included with document() function... Can't ivent any way
to pass this knowledge from XSLT procssor to Cocoon... :-(

>Another thing, why discourage something that's 100% part of the XSLT
>standard? Just because it does not work entirely correct (in my opinion it
>should not cache forever) does not mean you should discourage it...

The caching problems made me ask.. BTW: this was just an idea to think of.
We also use it ourselves in ;-)

>An example where I'd rather use the XSLT document() function instead of
>XInclude is in a project I'm working on. The user can create a configuration
>file which he/she can point to with a special tag in my taglibrary. The
>taglibrary reads the document with the document() function and will use it
>to specify some things in it's final JAVA file. I'd rather use document()
>overhere because I think something like this is a lot nicer then a XInclude
>approach will ever be, because people don't have to know anything about
><taglib:page config-file="...." optionabc="...">
> ....

Yes, lots of usefull things can be don by the "evil" document()..

Best regards, 
   Tagunov Anthony

To unsubscribe, e-mail:
For additional commands, email:

View raw message