Re: XPDL (was Re: XML is broken)

Subject: Re: XPDL (was Re: XML is broken)
From: "Simon St.Laurent" <simonstl@xxxxxxxxxxxx>
Date: Wed, 07 Apr 1999 10:46:36 -0400
At 09:11 AM 4/7/99 -0400, Ellitte Rusty Harold wrote:
[re: including pointers from documents to XPDL]
>I really wonder about this approach. It strikes me that the document should
>be the document, and all this differet information should come from outside
>of it. I really don't like the idea of embedding even a simple link to the
>XPDS stuff inside the document. I guess I'm prejudiced here based on my
>experience with style sheets where <?xml-stylesheet?> has caused me nothing
>but trouble. I hate having to edit my documents to apply a different style
>sheet, especially when I frequently want to switch between more than one
>style sheet for a single document.

I completely agree - the issues you're describing are precisely what drove
me to start writing XPDL.  (Okay, fine, there are a few others, but this is
critical.)  Section 3 of the XPDL draft (http://purl.oclc.org/NET/xpdl) now
has five different approaches for connecting XPDL descriptions to
documents, and I need to pare that down.  Ideally, none of this information
would live in the document itself, but rather in a separate resource with
control over that document.  Unfortunately, the infrastructures that
currently exist for this are pretty skeletal.

>What about using a standard naming convention for attaching XPDL info to
>documents? Or HTTP headers? Possibly but not necessarily backed up by a
>processing instruction, where the processing instruction has the lowest
>priority?  My fundamental problem here is that I view this as a sort of
>layered approach, and I don't want the lower layers (XML) to have any
>knowledge of the upper layers (XLL, XSL, XPDL, etc.)

I think assigning priorities is probably the best way to go.  HTTP headers
make a lot of sense to me - I just want something more meaningful than a
MIME type.  The PI provides functionality for situations where there isn't
any supporting infrastructure, just the file itself.

I definitely want this to be a layered approach, and plan to use it for my
layered parser implementations.  It's bothered me for a while that
_everything_ is being crammed into the document, from DOCTYPE to
<?xml-stylesheet?> to linking resources, and I'd like to see that change.
Self-identifying documents can be useful when there isn't much supporting
infrastructure, but can become a real hassle when they interfere with that
infrastructure.


Simon St.Laurent
XML: A Primer
Sharing Bandwidth / Cookies
http://www.simonstl.com


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list


Current Thread