W3C home > Mailing lists > Public > www-math@w3.org > July 2000

RE:Content markups, EBNF & DTD

From: Pankaj Kamthan <kamthan@cs.concordia.ca>
Date: Wed, 19 Jul 2000 21:31:09 -0400
To: quintana@lyon.objectif.fr
CC: www-math@w3.org
Message-ID: <39761E1D.7374.29D79D@localhost>
Gerald,

"The EBNF ... definition of MathML content markups seems to be 
far more detailed and precise than the DTD. What do you think of 
this? Is it normal ?"

This is good question for a [non-F]AQ. (David Carlisle has already 
answered
this on your posting of the same to MathML/Mozilla newgroup; 
here are some details.) 

The reason for having the EBNF for MathML Content Markup is to 
express constraints that can not be expressed in the DTD. 

As pointed out in reference 1. below, "DTD and BNF [and thus 
EBNF] are not equivalent. It is however possible to create an 
equivalent context-free (BNF) grammar from any XML DTD, where 
the terminals are start-tags, end-tags, and #PCDATA and the 
productions correspond to element types and content model 
positions. The reverse is not possible in general, so XML DTDs are 
(in a sense) a subset of BNF." So given a DTD, you could convert 
that into an equivalent set of EBNF productions; the converse may 
not always be the case.

There are reasons for providing both EBNF and the DTD. XML is 
based on SGML and the idea of providing a DTD comes from this 
heritage. (One of the major advantages of XML over SGML is that) 
the XML 1.0 Specification provides a EBNF for (among other 
objects) markup declarations (that constitute a DTD). EBNF 
notation is much more expressive than DTD. For example, an XML 
DTD does not provide information on the XML prolog but the EBNF 
does. Also, (as outlined in reference 2. below) the presence of 
EBNF naturally lends itself to a formal notation that is useful in 
parser, and in general, compiler design. However, using a DTD may 
be easier than using the information from the EBNF for authoring 
documents. 

If you only had the DTD, you could write a syntactically correct 
(according to the DTD) document. If you also had the associated 
specification (prose), then you would be able to infer the semantics 
of the elements and their attributes (among other objects) in the 
document. Furthermore, if you also had the EBNF, then you would 
know whether you should have the document correspond to XML 
1.0 (or some other version). So the knowledge of EBNF is 
necessary to author a document that conforms with the XML 1.0 
Specification.

Hope this helps.

Pankaj

1. SGML and XML grammars as context-free or not - By Joe English. 
http://www.oasis-open.org/cover/english-cfg.html

2. BNF and EBNF: What are they and how do they work? - By Lars Marius 
Garshol. 
http://www.stud.ifi.uio.no/~larsga/download/artikler/bnf.html 
Received on Wednesday, 19 July 2000 21:32:56 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 20 February 2010 06:12:49 GMT