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

-----Original Message-----
From: John Aldridge <john.aldridge@informatix.co.uk>
To: Tim Berners-Lee <timbl@w3.org>; xml-uri@w3.org <xml-uri@w3.org>
Date: Tuesday, June 20, 2000 7:07 AM
Subject: Re: Language = Namespace. was: How namespace names might be used


>At 15:58 19/06/00 -0400, Tim Berners-Lee wrote:
>
>>- a languge is (here) the set of names, their constraining syntax, and a
>>defined meaning for any combination of names which satisfies those
>>constraints on syntax.
>>
>>- a namespace corresponds to a language.  I know that some don't want this
>>model but honestly without it all work on XML should stop immediately and
be
>>restarted with a proper footing. What is XHTML? a Language! That is
actually
>>what the letter stands for. There is meaning in it.  The meaning is NOT
>>carried by out of band discussion, it is carried in the XHTML
specification.
>
>I'm still struggling to understand this vision.  In the example:
>
><a:a xmlns:a="http://a.com">
><b:b xmlns:b="http://b.com">
></b:b>
></a:a>
>
>Do we have:
>
>(a) A document written in one language using the set of names
>{http://a.com}a, {http://b.com}b.  This seems to be the natural
>interpretation of Tim B-Ls first paragraph.
>
>(b) A macaronic document written in two languages.  This is how I interpret
>the second paragraph.
>
>(c) A document which, although legal according to the namespace REC,
>doesn't conform to the vision of how the web will be built, and therefore
>doesn't need to be considered.  I sincerely hope that this is _not_ the
case.
>
>I _suspect_ that Tim-BL will answer (b), but I think that many of us
>(myself, certainly) have been thinking in terms of (a).


(a) and (b) both.
There is an abstract language for which
A and B are both sublanguages.   In that sense, (a).
But (b) is true.

["macaronic" adj. "Used to designate a burlesque form of verse in which
vernacular words are introduced into a Latin context with Latin terminations
and in Latin constructions. Also, applied to similar verse of which the
basis is Greek instead of Latin; and loosely to any form of verse in which
two or more languages are mingled together. Hence of language, style, etc.:
Resembling the mixed jargon of macaronic poetry".]

The xml-schma language actually allows the schema for B to refer to that for
A so that specific constraints on the superlanguage can also be expressed.
(Like "an SVG diagram can fit into XHTML where a block does")
It is in fact important to give some semantic link between teh
specifications
of languages which are expected to be combined, even if it is very simple.
For the first SMIL release, the relation was HTML was missing and this was
a problem.  (However,  if the languages are in fact RDF languages, then the
combination
can often be made (at the syntax level) without any extra information, as
they
share a grammar.)

>The important difference, it seems to me, is whether metadata is more
>usefully associated with a type (a) language or a type (b) language.  Will
>you ever need to make statements about an element which apply only in the
>context of another element from a different namespace, for example:
>
>     <b:b/> elements can be nested arbitrarily deeply when they
>     occur inside <a:a1/> elements, but may not be nested when
>     they occur in <a:a2/> elements.


I don't know whether you would want to make that specific statement,
but statemenst relating the languages are essential, and xml-schema
provides for them.


>Cheers,
>John
>

Received on Wednesday, 21 June 2000 17:38:55 UTC