RE: nodes or multiple runs?

Why are you creating temporary files? I have had several occasions to write transformation processing pipelines and I've always kept the output documents in memory and passed them on to the next stylesheet until the processing was done. I've done it with JavaScript, VB, PL/SQL, and Python so I know it isn't difficult to do.

Of course, you will still have to contend with different processors. You will have to write a tranformation section for each processor.
-- 
Charles Knell
cknell@xxxxxxxxxx - email



-----Original Message-----
From:     Paul Tremblay <phthenry@xxxxxxxxxxxxx>
Sent:     Wed, 11 Feb 2004 02:50:44 -0500
To:       xsl-list <xsl-list@xxxxxxxxxxxxxxxxxxxxxx>
Subject:   nodes or multiple runs?


I will be publishing a set of xslt stylesheets and want to make them
easy to use for everyone. My question is whether I should chain several
stylesheets together, or try to process the original document with just
one run, using temporary trees.

Right now, my transformation involves chaining together stylesheets. One
xslt stylesheet creates a document, and that document is processed with
an xslt stylesheet, and so on. 

This method requries that I create several temporary files. I can write
a python script makes the chaining together easy, but then I have to
contend with different processors. Saxon will require a different
processing command than Xalan. I will have to write a routine for each
processor, and that of course gets pretty involved.

However, I was reading Michael Kay's book, and I realized that I can
actually create temporary trees instead of making multiple runs. 

For example, I can group together all colors in the document and store
the result as a temporary tree in a variable. I can then number the
nodes in this temporary tree consecutively.

But not all processors can handle this method of using a temporary tree.
For example, xsltproc cannot. 

In addition, re-writing my stylesheets so that I can process everything
at once will take a lot of work. In fact, with my level of skill it may
not even be possible. The tricky part is that I have to number nested
lists according to attributes in the original XML document. A list that
is nested inside another list may require that it inherit the number for
its parent, or it may require that it not. The numbering style for
the child list may be deciaml, but the numbering style for the parent
may be Roman. The top level of each list must have a unique ID, and this
ID must be part of an attribute for the paragrphs. But children of lists
must not have ids. It is for this reason that I numbered all lists
first, and then on a subsequent run concatenated numbers from parent
lists when I needed to.

Is it okay to make a user process the document several times, or should
I try to make my xslt stylesheet a one run process?

Thanks

Paul

-- 

************************
*Paul Tremblay         *
*phthenry@xxxxxxxxxxxxx*
************************

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




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