forrest-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Steven Noels <stev...@outerthought.org>
Subject [Fwd: [xml-dev] ANN: RELAX NG to Compact Syntax Presentation in HTML (and XSLT Processor Implementer Challenge)]
Date Fri, 07 Feb 2003 14:27:46 GMT
Yum yum.

looks like I found the replacement for
http://forrestbot.cocoondev.org/sites/xml-forrest/document-v11.dtdx.html
once we have moved to RNG.

</Steven>

-------- Original Message --------
Subject: [xml-dev] ANN: RELAX NG to Compact Syntax Presentation in HTML
(and XSLT Processor Implementer Challenge)
Date: Fri, 7 Feb 2003 15:04:52 +0100
From: David Rosenborg <darolst@pantor.com>
To: <xml-dev@lists.xml.org>

Hi!

I've updated the stylesheets for RELAX NG XML syntax to compact syntax
conversion. New in the 1.4 release is a stylesheet for HTML presentation.
The stylesheet has the following features:

* Recursively converts input schemas to compact syntax, i.e., handles
include and externalRef
* Links from pattern references to definitions
* Links from pattern definitions to a generated index with back references

This stylesheet and stylesheets for plain text output are available in
the RngToRnc package at

http://www.pantor.com/download.html

Here is a small example:

http://www.pantor.com/rncexample.html

The new stylesheets use a two step conversion. First the main XML syntax
to compact
syntax conversion is handled by the RngToRncXml.xsl stylesheet. The
output is an
intermediate XML format. This format is then converted into either HTML or
plain text using the RngToRncHtml.xsl or RngToRncText.xsl stylesheets.
If your
XSLT processor provides a result tree fragment to node set conversion
function,
like exslt:node-set (), the two steps can be carried out up on a single
invocation.
See comments in the stylesheets for further details.

Caveats:

The RngToRnc stylesheets make use of two features that causes
portability problems:
advance namespace processing and r-t-f to node set conversion. Some of the
problems I've encountered was handled by disabling features depending on
the return value of system-property ('xsl:vendor'). In general, large
schemas are
more problematic than small ones. If you want to convert to plain text and
have problems with a large schema, you could try the RngToRncClassic.xsl
stylesheet which is a bit simpler.

XSLT Processor Implementer Challenge:

The challenge is to convert the office.rng schema in the following package

http://lists.oasis-open.org/archives/office/200301/zip00000.zip

to HTML using a single invocation of RngToRncHtml.xsl with the
retain-prefixes stylesheet parameter set to true.

I've successfully done this with a processor implemented in Java. It
took 12 seconds
on a 1.53GHz Athlon box, running RedHat 7.2. The time isn't really that
important
as long as it is within your patience limits ...

I've tested the same setup with Saxon 6.5.2 (never returns), Xalan-J 2.4
(never returns),
MSXSL 4.0 (on win2k, crashes) and xsltproc 20423/10019/701 (on RH8, 10 min,
partial output). 'Never returns' means that my patience run out after 15
min or so.

If you set retain-prefixes to false, Xalan works acceptable and MSXSL
works very well,
but xsltproc still produces partial output: it doesn't create the index
section. (I think the
problem in xsltproc is that it mixes up what is the current document
when finding the
root node and looking up key values.) Saxon works well for small schemas
in this
mode, but seems to have problems with the exslt:node-set () function for
large
schemas.

Admittedly, my namespace processing expressions aren't pretty, and could
maybe be
simplified to be kinder to the processor, but nonetheless, I expected
that the most popular
processors should do better than they apparently do.

Cheers,
David

------------------------------------------
David Rosenborg
Pantor Engineering AB



-- 
Steven Noels                            http://outerthought.org/
Outerthought - Open Source, Java & XML Competence Support Center
Read my weblog at            http://blogs.cocoondev.org/stevenn/
stevenn at outerthought.org                stevenn at apache.org


Mime
View raw message