Re: Language = Namespace. was: How namespace names might be used

-----Original Message-----
From: David Carlisle <david@dcarlisle.demon.co.uk>
To: timbl@w3.org <timbl@w3.org>
Cc: xml-uri@w3.org <xml-uri@w3.org>
Date: Monday, June 19, 2000 5:51 PM
Subject: Re: Language = Namespace. was: How namespace names might be used


>
>>  What is XHTML? a Language!
>
>XHTML 1.0 strict, XHTML 1.0 transitional, XHTML 1.0 frameset, XHTML
>1.1, XHTML Basic, and any languages built out of XHTML Modularisation
>are all languages specified by schema (dtd) and their respective
>specs.


In my model, XHTML 1.0 strict, XHTML 1.0 transitional, XHTML 1.0 frameset"
are each languages.   There is another language, XHTML 1.0, of which they
are all sublanguages

Defn:  A is a sublanguage of B if  taking a dcument written in A and
treating as though it were written in B will always leave its meaning
unaltered.
In XML, treating a document written in A as though it were written in B
means changing the namespace name from A to B but making other changes.

The community decided that the first three languages should have DTDs and
the
last one should have a Namespace ID.  They also decided that the namespace
would be a living object which can change at any time under W3C-HTML WG
control.  That is how I see what was done.  Alternative was of course to be
explicit about the three individual languages at the Namespace level
just as with the DTD.

That doesn't mean that another language definition can't give a permanent
namespace URI for a frozen version of the language.   That might lead to
less
confusion in the future.

>> The meaning is NOT
>> carried by out of band discussion, it is carried in the XHTML
specification.
>
>specification_s_ not specification.
>
>Many (related) languages, all using elements from the same namespace,


Many langages related by being sublanguages of the same language.

>> It is really important that this
>> specificatoin can refer to the language (xml-schema as of today) in which
it
>> is written.
>
>well in that case it had better reference by schemaLocation to the
>dated URI for the schema for schemas, as it is similarly important
>that any future updates to the schema for schemas use the same
>namespace.

Why?  A change to the schema spec in fact occured.
Making it "the same" namespace was great for everyone who
didn't have to alter their documents and could still reference the
lastest schema spc, but it actually broke the XML DSig schema.
I don't like that.
I don't think the solution is to use the schema location.
The schema spec could have been changed to break the
DSig schema quietyly -- just changing the meaning of
elements while keeping the schema-for-schemas intact!
To make the DSig schema fully defined, it has to
be able to bind the terms it uses to their meaning
(as well as the syntax constraints).  Then it is a grounded
document.  This has to be something the namespace does
as to have different identifiers for the tokesn, the syntax, and
the  meaning would lead to so much confudion and ambiguity.

The model we are dioscussing with namespace-language
is simple and does what we want.

Tim

Received on Wednesday, 21 June 2000 17:26:25 UTC