Lists Home |
Date Index |
You are both right. To work at this problem, one has to
start with XML 1.0. Then the infoset, then schematization.
The notion of basing a system on well-formed strings doesn't
begin with XML. The notion that a formal system must have
passive meanings doesn't begin with relational databases.
One might want to look at what is and is not formally
expressible and by what forms (avoid the interpretation
initially and just deal with the axioms and theorems).
It isn't useful to attempt to prove XML is as expressive
as a relational database. It can be useful to determine
which kinds of theorems produce isomorphic meanings and
those that can't. For that, you XSLT is useful
For example, the object-oriented impedance mismatch is one.
The rules for theorem creating don't enable attributes to
have elements. So a transformation from object languages
where fields have object can't be done by transforming fields
into attributes. One has to transform the fields into elements
at the higher level of the 'names'. The meaningfulness
is determined by the isomorphism. XML can express a programming
language, but the forms are restricted by the formal properties
of XML 1.0.
<aside>One of the reasons namespaces
get a sense of revulsion is that they are a foreign system
grafted into the grammar.</aside>
XML is a formal system with provable properties given that
XML 1.0 provides the rules and the syntax. I *believe* it
is more expressive than a relational database
but that this forces schematization to reduce the Boltzman
entropy problem and in that way, is no different than
the relational system. Rules is rules.
From: Simon St.Laurent [mailto:email@example.com]
firstname.lastname@example.org (Chiusano Joseph) writes:
>one set might be xml with tags only - no attributes; another might be
>xml that is constrained to two levels; etc
>What value would there be to labeling these types of XML (e.g.
>"attribute-less" XML, "two-level" XML)? This seems to me to be
>something that can be better covered by an XML schema design (if we
>are talking about schemas here) than a broad classification. IOW, an
>organization/agency may decide for whatever reason that they want to
>avoid the use of attributes. It also appears to me that the
>combinations/permutations of the different aspects here can become
>quite extensive and perhaps unmaintainable.
I'm not sure I'd knock this so quickly - there's something interesting
here. I suspect the math around this is already done, in the hedge
automata work that's underneath RELAX NG's pattern approach. Exploring
this formally might not be as difficult as you suggest.
That said, I don't speak the math well enough to be able to say that for
The xml-dev list is sponsored by XML.org <http://www.xml.org>, an
initiative of OASIS <http://www.oasis-open.org>
The list archives are at http://lists.xml.org/archives/xml-dev/
To subscribe or unsubscribe from this list use the subscription