Re: Link issues

In message <3.0.32.19970518111005.00b18a30@pop.intergate.bc.ca> Tim Bray writes:
> People have reacted quite positively to the contents of the XML-link
> draft.  However, the work is *very* far from finished, and July 1st
> is galloping toward us.

This is a first response, and written from the point of view of an 
implementer.  (Much of XML-LINK is already in JUMBO, but rather kludgy in
places, because when you don't know where you're going, it's tricky to 
plan :-).  However I'm doing a rewrite and things should be a lot simpler.
There is no intention to have a fully compliant JUMBO by July 1, but I hope
it's possible to make fairly accurate pronouncements on the likelihood of
the spec (a) being interpreted identically by implementers and (b) being
implemntable at a few weeks of MCSG97.)

As a baseline I am even more convinced that we need an application-independent
post-parser (pre-application).  There are a lot of implied semantics in 
XML-LINK and a considerable amount of syntax which (IMO) requires validation.
I've posted this before so won't elaborate, but an example is that an
EXTENDED element can only contain #PCDATA and LOCATOR.  (Agreed, the DTD 
can be written to require this, BUT we have to cater for WF documents as
well - do they have to obey this syntactic requirement).  My opinion is
that there are some tricky issues about how WF documents should/may/must
behave under XML-LINK and these need addressing.  The first decision is
whether they require the same level of syntactic and semantic validation as
validatable documents - I have no strong opinion (yet).

I also feel that XML-LINK is so powerful that it opens up a wealth of 
new applications.  Therefore I think it's important that (say) the design
of XLG allows for a 'document' to be a generic node in any data structure
and that limiting its use by analogy with printed documents is too 
restrictive.

> 
> The next few messages describe a bunch of open issues in XML-Link
> that we need input on if we are to make our deadline.

I have no views on the syntax of L1 (HERE).  JUMBO can be hacked to 
accommodate it whatever the solution.  However I suspect that XML-LINK
will start to generate a set of KEYWORDs, reserved words or whatever, and 
that this will have to be addressed soonish.

I shall comment on how JUMBO sees issues L2-6 in separate postings.  Most
are paryially implemented.

In general I think XML-LINK is implementable.  Until we start to see 
implementations widely I suspect there may be hidden problems - I've flagged
the ones I can see, of course.  That may mean another iteration and I
suspect we have to face up to that possibility.

	P.


-- 
Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences
http://www.vsms.nottingham.ac.uk/

Received on Sunday, 18 May 1997 09:06:24 UTC