Re: SD5 - Namespaces

Peter Murray-Rust wrote:
> 
> > Unless I missed the point, you are saying you want to divide a GI by a
> > colon to provide and application with a morphology for a namespace. 
> 
> Yes.  I want to write things like <CML:VAR> and <MathML:VAR> in the same
> document.  My reading of the spec was that these are Names [5] and that the
> only currently allowed chars in names are Letters, Digits, '.', '_' and '-'.

I had to think about it.  Old apps that took this approach
post-processed 
(at least after the SGML parser was done).  You want the XML parser to 
help.  Ok.  That makes sense. 

> So I would like ':' added to the list of MiscName.  I need the ERB's
> blessing for that, because parsers are written to the current spec.  I can't
> use an SGML declaration (and wouldn't even if I could, because my community
> would rebel!).

That argument is strange to me and always will be.  We are waaaay past 
the average bear at this point anyway and SGML is only hard if one 
has to write the parser.  I suspect before we are done, 
we will have SGML Lang and XML app for everything else.  But hey, 
eternal optimism here... :-)

> Whether it goes further and there is a formal definition of name as
> Name ::= (Namespace ':')? ShortName
> or something similar I leave to the ERB.

At least, keep up the discussion.  The ERB can come down to a 
simple majority and nothing guarantees it will be the most 
technical of the collective.  That is why this chorus is 
here.
 
> Whilst on this topic, I read MartinB's post about namespace.  Obviously if
> a WG8 solution is going to emerge it will be thought out to apply beyond the
> confines of XML.  

Yes. OTH, if developers really need this now, we have to address it now. 
That is good.  If we find a workable and acceptable method, then 
SGML WG8 (most of whom are here lurking somewhere) gets to see it 
worked out, tested, shredded and ammended. Then if it still makes sense, 
they have a very firm basis for adoption.  I think that is a good
process.

> The colonic approach 

high or low?

> ..is implementable today with no changes other than [3],

Sure.  Now, can we say in short sentences what the requirement(s) is?
That would probably help the authors of the spec as much as anything.

> As Len and others have pointed out, we have to address the namespace or
> DOCTYPE becomes as useful as it is in HTML :-(.  We are not addressing the
> people who produce tag soup like
> 
> <Q> What did Queen Victoria say?</Q>
> <A> <Q>We are not amused</Q>, see <A HREF="home.html"> my Home Page</A></A>
> 
> but those who already have DTDs and where both the syntax and the semantics
> have to be precisely identified.

Yep.  It's the other part of XML and it needs work.  The folks to do
that are certainly here.  Most of my day is now spent looking at
commercial 
relational database apps, so I can only sit and learn from you at the
moment.

len

Received on Thursday, 22 May 1997 20:52:06 UTC