xml-dev - Re: Does DTD validation work with namespaces?

[Norman Walsh <ndw@nwalsh.com>]

/ "Winchel 'Todd' Vincent, III" <winchel@mindspring.com> was heard to say:
| Amy's vision is my own.  For example . . .
| 
| Within a "Legal" set of namespaces, (Court Filing, Contract, Transcript), if
| the same legal industry consortium defines the namespaces, then they
| *should* work together as Norman suggests.

It seems that in some respects my vision and Amy's are different, so
I'm not sure how to intepret what you've just said.

| However, I agree with Amy and disagree with Norman in the following respect
[...]
| Contract DTD into their website mark-up, or an Amazon invoice, or standard
| prospectus mark-up with an independently created DTD defined by some other
| organization, or anything else that the legal industry consortium does not
| know about (and did not consider when it created Contract DTD), then this
| would be perfectly *appropriate* and, indeed, exactly what the goal should
| be.

If I follow, what you're saying is that there are contexts in which you'd
like to make random mixtures of markup. Embedding contract markup, for
example, in a web page.

I agree that that is perfectly reasonable, but I don't think that it
is "valid" in the traditional sense. Maybe we need some new definition
of valid for this case, "mixed-namespace-island validity" or
something.

| Norman, you missed the point when you replied:
| 
| / "Winchel 'Todd' Vincent, III" <winchel@mindspring.com> was heard to say:
| | It seems to me that URIs would be the right answer if there were a
| | one-to-one relationship between URI and namespace prefixes, rather than a
| | one-to-many relationship (i.e., unique prefixes via a fixed association
| with
| | a URI).
| 
| <Norman Date="2000.07.25" Subject="Re: Question About Namespaces and DTDs">
| If there was a one-to-one relationship, then the URIs would be
| unnecessary, the prefixes would be enough. But any proposal based on
| the notion of standard prefixes amounts to little more than adding
| another name character to XML. And that's just not enough.
| </Norman>
| 
| You took the above paragraph out of context.  In the previous email, I also
| wrote:
| 
| <ToddVincent Date="2000.07.25" Subject="Re: Question About Namespaces and
| DTDs">
| I have been thinking that it would also be nice if there were some
| requirement (perhaps an optional feature in parsers) that allowed one to
| fetch the schema/DTD at the end of the namespace URI.  If there were at
| least a moral responsibity on the owner to keep the schema/DTD at the end of
| the URI the same (or tell people when it changed) (or a perhaps a
| responsibility on the user to hash/sign it, so you know if it has changed),
| then namespaces would be tied to a vocabulary that gave them meaning (now
| and in the future), rather than simply being a means of avoiding technical
| element collision.  This is what I thought "semantic web" meant and was
| disappointed to find it didn't.
| </ToddVincent>
| 
| My understanding is that URI's can't be parsed, but prefixes can.  Prefixes
| don't have a well-defined means of pointing to web resources.  So, prefixes
| and URIs are very different.  A one-to-one mapping of them makes a lot of
| sense if you want to use the prefixes on elements you want to parse and also
| use the associated URI to point to a DTD or Schema that gives the elements
| some meaning.  (Please correct me if I'm making any technical mistakes
| here.)

If you aren't suggesting that "html:" always mean
http://whateverthexhtmluriis/, then I'm not sure I follow.

| If well-formed Contract XML (defined by X consortium), embedded within
| Invoice XML (defined by Y consortium) points to a standard, fixed,
| well-known DTD or Schema, which in turn references a specification (or
| another Schema perhaps) that explains it, then I think Norman's conceptual
| problem begins to go away and we start to have a "semantic web."

I understand that there are several possible ways to interpret
validity.  For my own, personal use, I'd only be happy with the above
scenario if there is a schema Z that references X and Y so that it
explicitly indicates in what content models the various mixtures are
allowed.

But I am by no means saying that I think my understanding of validity
is (or should be) universal.

| There are really two different, but related issues here.  (1) Namespaces
| don't work as a technical matter if you want to validate DTDs

They work OK as long as you are in a closed system and can always use
a fixed set of prefixes *and* you have my draconian view of validity.

Extending DTDs to work outside this environment, by means of a PI for
example that mapped prefixes to URIs in the DTD, would make a parser
that accepted documents that were not XML 1.0 valid but were XML
1.0+Namespaces valid. It'd be an interesting exercise.

If you want a more liberal view of validity, you have to use ANY
content models. I really don't think you can get the "these elements
from my schema, or any element from these three other namespaces, but
nothing else" semantics from DTDs.

| (2) Namespaces
| don't work very well conceptually, even if they work technically, whether
| you use DTDs or Schemas, because there is no requirement or even expectation
| of uniqueness (unless you work within your own closed little world), so when
| you start mixing and matching elements from different schemas that you don't
| know about, you get a jumbled mess of elements that may *technically* work,
| but has no meaning (or at least a meaning that is very uncertain and that is
| subject to change at anytime).

Uniqueness is garaunteed by URIs. You don't get prefix uniqueness but
you don't need that if you have the mapping to URIs.

                                        Be seeing you,
                                          norm

-- 
Norman.Walsh@East.Sun.COM | I have animal magnetism. When I go outside,
XML Technology Center     | squirrels stick to my clothes.
Sun Microsystems, Inc.    |