Re: Validation of MathML

Hi Simon,

> I get the impression that most MathML users rely on the validation by their
> authoring tool. Is that right? 

I think that is largely the case.  Naturally some of us have had to
deal with validating hand-authored MathML, etc.  But mostly it's been
handled in an ad hoc way I suspect.

> So we seem to be adding a new aspect to MathML validation:
> validating documents containing MathML from various sources,
> independently of the creation tools.
> 
> While I was working on this, I started to wonder why the DTD is formulated
> in such a loose manner. At first sight several of the requirements listed in
> section 3.1.3.2 could have been enforced in the DTD (that would only be for
> presentation MathML; I am not sure whether the same would be true for
> content MathML). Or are there hidden problems, which I will discover when I
> try to work out this idea?

For MathML 1, we started down the road of making a tighter DTD, but
soon came to the conclusion that a) it would make it much more
complicated, and b) still be a long way from truly validating all the
MathML rules since that would require parsing attribute values, etc.
Also, I think you will find some hidden problems in the content MathML
area, where the validation rules require a stronger notion of context
than is possible with a DTD.  

So we decided to just acknowledge the limitations and go with a fairly
minimal DTD.  I'm sure that you can beef it up quite a bit if you
want.  

However, another factor in the decision to go with a minimal DTD was
the sense that schemas were coming down the road, and they were
supposed to be more powerful for validation.  Now in fact, schemas
have been slow to arrive, and there are still serious issues with
them.  But we have now produced a schema for MathML 2 that does much
more validation than the DTD.  You can get it at

   http://www.w3.org/Math/XMLSchema/

It's should still be considered a work in progress, but you may find
it useful.

--Robert

------------------------------------------------------------------
Dr. Robert Miner                                RobertM@dessci.com
MathML 2.0 Specification Co-editor                    651-223-2883
Design Science, Inc.   "How Science Communicates"   www.dessci.com
------------------------------------------------------------------

Received on Thursday, 27 February 2003 11:04:07 UTC