[SCL] some explanation

[Pat Hayes]

>Steve,
>
>That is the fundamental point:
>
>SW> Or to be more general:  don't let a specific
>>  serialization format influence the abstract syntax
>>  (as can happen with XML-based stds).
>
>As various people have noted, an anti-XML backlash
>is growing strong enough that even the W3C fanatics
>are beginning to feel the pressure.
>
>We should *not* allow anybody's pet notation to
>pervert the semantics.  For political reasons, we
>can emphasize one or another concrete syntax in
>different documents, but the core must be clean.

What does 'clean' mean?

Stating an abstract syntax is inherently awkward, since you can't 
write down an abstraction; you have to write down something concrete. 
So what you finish up doing, if you try to do it in detail and 
precisely, is writing a concrete description of an abstraction, ie a 
description of an algebraic, mathematical syntactic structure. And to 
state the required meta-properties, such as 'for any term T, ... ' in 
the surrounding text, you typically need to have names for the 
various parts of the expression being described (the third term in 
the sequence which is the body of the atomic sentence..). By the time 
you have invented all this notation, you have something that is 
already as complicated as a fully XML-decorated text, and less 
readable. And this really is a 'pet notation', since there are no 
agreed-on standards or even mathematical notations or terminology for 
doing this meta-syntax description (recall, I had to point the SCL 
group at McCarthy's old papers to even explain the idea to them.)

So, why not use XML to do this? Annotating text with markup is pretty 
much a definition of XML. And then the XML description can be the 
abstract syntactic description, with the syntax categories in the 
tags, and one can use the same DTD for the XCL concrete syntax, with 
actual CL stuff inside the items, and for the 'abstract' syntax, but 
now with metasyntactic variables inside the items. The result both 
saves a completely unnecessary step in the overall specification - 
for your 'clean' abstract core is in fact not the slightest use or 
interest to anyone but logicians - and also uses a fully formalized, 
widely used, and machine-processable notation to do exactly the kind 
of job it was designed to do.

However, mea culpa for not having adequately explained this idea to 
the group in time for y'all to consider it rationally.

Pat


>
>John
>_______________________________________________
>SCL mailing list
>SCL at philebus.tamu.edu
>http://philebus.tamu.edu/mailman/listinfo/scl


-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
phayes at ihmc.us       http://www.ihmc.us/users/phayes