Re: SD5 - Namespaces [fmt]

[Tim Bray]
> >The demand for namespace machinery is sudden but, it seems, overwhelming.
> >The demand can't be ignored.  Once again, it should go in a separate
> >volume in the lang/link/display series.  Once again, it should be done
> >in an 8879-compatible way.  Once again, it can't be done properly by
> >July 1.

On Mon, 19 May 1997, Paul Grosso wrote:

> I don't have much to say about namespaces yet, except that I pretty
> much agree with what Tim says, 

Me too! :-)

> and I don't think subdoc (at least as currently defines in 8879) is 
> the answer. I think what we need will require something beyond what 8879
> offers us now, 

I'm still somewhat puzzled by the SUBDOC suggestion, since "shifting" into
a subdoc is supposed to establish a new parsing context (see p.90 in the
Handbook on Annex C.3.2), whereas from what I can tell the closest thing
to simultaneous namespaces -- which is my undertsanding of SD5 -- in ISO
8879 is CONCUR. However, having never used this, I have no idea how big a
can of worms I just opened here:-) [If I should be shot, could someone
please do that ASAP?:-)]

> >One detail point.  Using the dot '.' to separate schema & name may 
> >have a nice O-O feel but lots of element types already have dots in
> >them.  We should change XML-lang to add ':' or some other handy
> >character to the name repertoire,  but forbid its use except for
> >this purpose.  Then people can use any legal XML name for their
> >schematized element types. -T.

Nevertheless, drawing on simply the syntax of CONCUR-ed tags, the schema
could be placed in parentheses, viz <(schema)tag>. In fact, production
[28] in ISO 8879 says that's really a name group, so multiple schema
could also be placed within the parens, appropriately separated.

Complicates the parsing somewhat though (and calls for more than a
cosmetic change to the -lang spec compared with Tim's suggestion.)


Arjun

Received on Monday, 19 May 1997 22:58:18 UTC