Re: [XPath] A.2.2 Parsing note

scott_boag@us.ibm.com wrote:
> 
> One thing I am exploring is expressing these tables as a mini-bnf (you
> yourself may have suggested this in the past),

Yup, two years ago:
http://lists.w3.org/Archives/Public/www-xml-query-comments/2002Jan/0002.html

> either to do validation of the tables,

Do you mean an algorithmic validation, or just "see if we can spot any
errors"?

> or to propose to the WG as a replacement for the tables.

Which brings up the question I raised two years ago: "Why do you need
two CFGs to define the language?"


> Another possibility is to go ahead and try to generate the tables
> from the bnf+some hints.

It's easy enough (even without hints) to generate verious kinds of
push-down automata from a context-free grammar: that's typically what
parser-generators do. Of course, if you did this with the A.1 EBNF, the
resulting PDA wouldn't be the same as the one in A.2.2, because it would
recognize the actual A.1 language, rather than a superset of it (as the
A.2.2 machine does).

If you could explain what it is about the ad hoc A.2.2 machine that
makes it preferable to a conventionally-derived PDA, we could maybe make
some progress toward eliminating section A.2.2.

-Michael Dyck

Received on Sunday, 8 February 2004 17:06:03 UTC