W3C home > Mailing lists > Public > xmlschema-dev@w3.org > December 2000

Re: What does 'non-deterministic content model for type' mean?

From: Henry S. Thompson <ht@cogsci.ed.ac.uk>
Date: 20 Dec 2000 16:09:37 +0000
To: Michael.Burns@sas.com
Cc: xmlschema-dev@w3.org
Message-ID: <f5bofy713cu.fsf@cogsci.ed.ac.uk>
Michael Burns <Michael.Burns@sas.com> writes:

> I have a fairly complex schema (2937 lines, 822 elements, 617
> complexTypes) that xsv
> is reporting a schema error on:
> 
> <schemaError char='1' line='385' phase='instance'
> resource='file:///u:\mit\mdf\m
> c\test\mitjin\scratch\SASOMI.xsd'>non-deterministic content model for
> type Metad
> ataType:
> {http://www.sas.com/SASOMI}:CubeRegion/{http://www.sas.com/SASOMI}:Cube
> Region</schemaError>

It means that in the content model for MetadataType there are two
alternative routes through the model to a CubeRegion element, e.g.

((a?,CubeRegion)|(b?,CubeRegion)) or (CubeRegion*,a?,CubeRegion*)

This is not allowed, just as it wasn't in SGML and XML.

> What does this mean?
> FWIW, xsv does eventually properly evaluate the instance and detect
> the one
> known irregularity in the content.  (shown below).
> It just takes it 12 HOURS on my 400Mhz PIII !

Most recent release (XSV 1.173.2.6/1.82 of 2000/10/31 16:30:47) should 
give order of magnitude better performance.

> The actual instance is very small (26 lines).  I suspect most of this
> time is in
> digesting the schema.  It is very complex and has some circular types
> of element
> associations ( element A can optionally contain a B, and B can
> optionally contain
> an A).  Is this why it takes so long?

No, it's very large FSMs that cause the problem.

ht
-- 
  Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh
          W3C Fellow 1999--2001, part-time member of W3C Team
     2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
	    Fax: (44) 131 650-4587, e-mail: ht@cogsci.ed.ac.uk
		     URL: http://www.ltg.ed.ac.uk/~ht/
Received on Wednesday, 20 December 2000 11:09:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 11 January 2011 00:14:19 GMT