Re: namespace viz validation [was RE: DSIGS]

At 2:03 PM -0500 6/18/97, Paul Grosso wrote:
>At 14:39 1997 06 18 -0500, David G. Durand wrote:
>>At 8:38 PM -0500 5/23/97, Andrew Layman wrote:
>>>Oh.  I agree that we don't want to preclude validation.  Just that
>>>validation and namespaces are separate issues.
>>
>>As part of my review of namespace blather I came upon this anti-gem.
>>
>>Since validation is not possible in the presence of arbitrary included
>>markup unless that markup is part of a DTD, the GI-hacking namespace
>>proposals are all intimately related to the question of validation. Wishing
>>will not make the problem go away, so we need to address it if we adopt any
>>such proposal.
>
>I'm having trouble finding the technical arguments in this posting, but
>I wanted to say that, while it's always true that all issues are
>inter-related to varying degrees, I find it helpful to separate namespaces
>and validation for the most part.

I agree that I'm getting testy, since I've posted 3 previous postings with
the argument clearly spelled out, I may have said to little. Apologies for
any rudeness, but the point still stands. I will repeat the argument again.

If you allow "importation" of elements from another "namespace", you need a
way to validate that this is legal according to the DTD.

But the reason you want to do this is because the element is _not_ legal in
the DTD.

Therefore the validation of multi-namespace documents needs to be defined.

So we have a technical argument that you can't have namespace proposals of
the form advocated without defining some new kind of validation behavior,
or without creating a blanket exception from DTD conformance requirements
for validating parsers.

If you want to do something other than such partial DTD importation, like
just attach unique IDs to elements of some sort, then you don't need
namespaces, just a convention about how to use attributes.

Clear now?

>For me, the basic namespace issue comes down merely to figuring out
>some way of (1) "uniquifying" names in an instance so I can tell that
>the <a> element in one part of my document (the HTML namespace) is
>different from the <a> element in another part (the AAP equation
>markup namespace), and at the same time

Attach an attribute. You can use Notation or ENTITY valued attributes to
associate an FPI with it, if you want.

(2) grouping names so that,
>for example, I know which style sheet to use to find information about
>this name.  None of this requires validation.

Each stylesheet looks at the attribute corresponding to its architecture. I
thought this was well-understood procedure by now.

If you use Attribute values to control processing, you can decouple or
tightly couple DTD and semantics at your pleasure.

>So, while I agree that we need to address how validation could work
>with multiple namespaces--just like we need to address how validation
>works with any issue we may raise with respect to XML--I see the basics
>of the namespace issue as potentially independent of validation.

As long as you stay away from allowing things in one DTD that are not
declared in that DTD.

I don't see a strong emphasis on this foundational point, and you're the
first person to respond to my posts on the issue, so I may be getting a bit
testy.

_________________________________________
David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________

Received on Wednesday, 18 June 1997 15:38:57 UTC