RE: Proposed restatement of syntax-based interoperability princip le ( was RE: Action item on syntax-based interoperability)

From: Bill de hOra []

>> I don't believe syntax is fundamental in the sense that applications 
>> have to share a syntax to interoperate.  Again, there are too many 
>> existence proofs that disprove this is necessary.  

>I'd like to hear about them.

Start with VRML97 and X3D.  As long as the abstract model is 
solid and the encodings support that, you can have multiple 
syntaxes and interoperate successfully.  What you lose is 
easy integration with other languages, and anyone who opens 
an ASP page and looks at the multiple syntaxes understands 
both the solutions and the costs.  On the other hand, if you 
want true interoperation between the attractors of behavioral 
fidelity and rendering fidelity, the syntax is just a throttle 
on performance.  The Classic VRML encoding is waaaaay faster 
than the XML encoding and there is no impedance mismatch because 
of the structures.  RNC advocates say they need the alternate 
encoding too for convenience (easier to teach; easier to read), 
so how fundamental is the syntax compared to the functional 
model?  Not much.

>> o Convenient - this is a piece of the architecture of any 
>> information system for which it is not difficult to get buy in.
>> Most designer, manager, data owner or supplier will understand 
>> why this makes life more convenient.

>If that were the case, surely this thread would be unneccessary?

If the architecture is based on declaring that the 'web is an 
information space' and the document can get away with that bit of 'hand 
leaves arm and flies around room; miracle occurs here', there 
is no end to the unnecessary threads that can be closed with a 
bit of religious dogma.  And that is all the 'syntax is fundamental' 
statement is:  dogma.  It is true in practice but not for any 
reasons more fundamental than 'cheap and convenient', and those 
reasons are sufficient.

> Cheap and convenient may not be words that resonate loftily in 
> an architectural tome, but they make sense, even, common sense. 
> And that has value.  IMO, an architecture document of the 
> primacy of this one should not include statements of religion, 

>Sorry, I'm with Tim Bray. 

Don't apologize.  He is right about some things.  He is out 
to lunch on others.  Welcome to the chicken coop.

>Syntax has been fundamental to web based interop in my experience and

Why?  Because when a 'system' has that many users, cheap convenient 
pet tricks recoup very large costs.  Those costs come in many forms 
including the potential wrangling over the holy brackets (Shall We 
Make Curly or Pointy Holy?).  But even then, the impedance mismatch 
that a syntax specification with a namespace and a structure can 
cause create very real headaches.  XML is the winner of the 'pick 
one' contest.  Syntax can be very important. Is it important for 
everyone to pick one?  No, but it is cheaper and convenient.

>Whether or not we have a 
>good theory on the matter to back that up is another matter - I 
>suspect the current notion of "partial understanding" doing the 
>rounds in semweb circles might have something to offer.

You'll have to explain.

>On the other hand it seems that saying syntax (or an indexical) is 
>fundamental, or more fundamental to interoperation than either 
>semantics or shared models, is at best troubling, at worst heresy. 
>I'll close by noting  that no-one is saying that syntax is sufficient.

Because I never have been able to get this coop to do more than cluck 
when asked for a definition of 'interoperation', I have to agree with 
the last statement.  It isn't sufficient.  But I have to add, nor 
is it fundamental.  Very useful, cheap, convenient, yes, but nothing 
about the web except scaling costs make it necessary, and necessity is 
my definition of fundamental.


Received on Monday, 27 October 2003 13:39:00 UTC