[
Lists Home |
Date Index |
Thread Index
]
- From: "David Brownell" <david-b@pacbell.net>
- To: "Richard Tobin" <richard@cogsci.ed.ac.uk>
- Date: Thu, 18 May 2000 16:45:50 +0200
> > So after reading 'PUBLIC' you have no idea if you are looking for
> > "S PubidLiteral S SystemLiteral " or just "S PubidLiteral"
> >
> > What am i missing ?
>
> Nothing, but what's the problem? It's ambiguous (in the content model
> sense of non-deterministic), but the productions in the spec are not
> intended to correspond that directly to an implementation.
Actually I think the XML spec would have been substantially improved,
in the technical sense, were it to have been directly validated by an
implementation -- using only the standardized productions.
It's pretty common for language specs to ensure that their grammars
can easily be handled by parser generators -- commonly they'll be
done as LALR(1) [yacc/bison/...] or somesuch. That level of precision
helps eliminate interoperation difficulties of various kinds ... and in
this case would surely have turned up some of the layering issues
with respect to parameter entity processing. (Well, I look at them
as layering issues -- the "%PE-name;" constructs apply somewhat before
the grammar rules, but how and where to apply them depends on some
contextual rules that have taken a long time to start getting clear.)
It's not inherently a problem if the specification grammar isn't
handled by standard parser generators, but in this case I think
it would have made a difference. That "implement in a week" goal
would likely have been achievable if the spec had addressed such
issues earlier.
- Dave
***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************
|
