RE: Validation of MathML

Pepping, Simon (ELS) [mailto:S.Pepping@elsevier.nl] wrote on 19 March 2003
16:04:
> David Carlisle [mailto:davidc@nag.co.uk] wrote on 27 February 2003 21:01:
> > Simon,
> > 
> > Others have mentioned the schema, but assuming that you want a dtd
> > version to fit into a larger dtd based document type, have 
> you noticed
> > that the current version has an option to turn on stricter checking
> > just set
> >    <!ENTITY % MathMLstrict "INCLUDE">
> > before including the mathml dtd into your dtd.
> > 
> > there is some documentation of this feature in appendix a 
> of the draft
> > version of mathml2 2nd edition.
> > 
> > I'd be interested to know if
> > a) these extra dtd based tests would be useful to you
> 
> I had constructed a number of files with errors that were not 
> caught by DTD
> parsing. With this new version all errors were caught by the 
> parser, except
> for an error with the number of post- and prescripts in 
> mmultiscripts. This
> is very useful; I am going to use this DTD in our validation.

It might be useful for other users to report that at first I had serious
problems parsing files with this DTD (XHTML + MATHML) with Xerces on Windows
2000. This turned out to be due to the fact that Sun's 1.4.x java kits
silently fall back on the version of Xerces shipped with it. With that
version of Xerces this DTD sends java into a loop, consuming 100% of the
CPU. The current version of Xerces has no problem with this DTD. 

On that matter, I have not been able to parse any XHTML + MATHML file with
libxml (xmllint), with any version of the DTD. It reports the weirdest kinds
of error.

Regards, Simon

Simon Pepping
DTD Development and Maintenance
Elsevier
s.pepping@elsevier.com
www.elsevier.com/locate/sgml

Received on Thursday, 20 March 2003 04:09:25 UTC