Re: Meaning of normative references [was: Update on namespaces]

Dan Connolly wrote:
> 
> David put forth a conjecture:
> 
>         IF XML cites ISO8879 normatively,
>         THEN implementors of XML are implementors
>                 of ISO8879, and hence may need to consult
>                 that spec.
> 
> You claimed the conjecture is false, and I have asked
> you to defend your position. 

That was not the conjecture made. I said that it was false 
that a technically complete specification could not 
be written which includes a normative reference.

> I'll ask one more time. If
> I get rhetoric in return this time, I'll just dismiss it;
> I encourage others to do the same.

Threats are rhetoric too.  I suspect those whom 
you could encourage have less influence than you 
require to have an effect.

> Sometimes I find your rhetoric entertaining. 

Tit.

And sometimes I find your attempts to use logic 
to disguise politics funny.  

Tat.

> But this
> is one of the few black and white issues around, and I
> don't appreciate your rhetorical answer.

So?
 
> > Consequence arises from action not declaration.
> > Otherwise, the word would be the thing.

What followed was simple enough.  A normative 
reference does not preclude the editors from 
writing a technically complete specification 
that requires the implementor to read ISO 8879.  
The consequence arises from the action;  if the 
specification is incomplete, the fault lies with the 
authors.

The idea of XML 1.0 (the number) is to enable that 
with a smaller scope.  If by increasing that scope, 
they make their jobs undoable, that is an action 
that has consequence.  They can't do their jobs. 
That is straightforward and not rhetorical.

Is XML a subset of SGML?  If so, by what measure 
or action can one ensure it remains so?

Conjecture:  You want a non-normative 
reference because essentially, you don't want XML 
to be SGML.  Is that the case? 

A normative reference prevents that in definition 
if not in practice.  The lesson learned from 
HTML and HTTP was that no definition can 
prevent bad practice.  Usually, conformance 
testing is the best compromise for enforcement, 
but in the end, justice under law is a practice 
itself, and this tests men, not software.

Rhetoric?  Well, then most law is rhetoric.  
Read a few Supreme Court decisions.  It comes 
down to authoritative opinions, not proofs.

The other benefit of scoped actions is that they 
enable other actions to occur with some rough 
synchronization.  In this case, the desire of many 
is that as XML design proceeds, SGML ammendments 
also proceed to enable these to remain 
harmonized and mutually reinforcing.  Is this 
not a good goal?  Is this not a goal of the W3C?

It is an issue of authority.  W3C 
recommendations or ISO 8879?
I had hoped this was not at issue.

len

Received on Tuesday, 24 June 1997 21:30:46 UTC