RE: Type and Structure Re: ASN.1 and XML

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Sorta.  Even more ancient works of lore have some 
notions of the sepration of syntax and semantic if 
only as separate stacks.

"In any PL implementation, that are actually at least two 
stacks which work in parallel.  One, the syntax stack SYN.. 
The second, called the semantic stack SEM, holds the "meaning" 
of the symbols in the first stack.  That is, SEM(i) holds the 
"meaning" or "semantics" of the symbol in SYN(i).  ... 

In any implementation of PL, these is also an associated 
semantic language, in which the semantic routines are 
programmed. This may be ... it really doesn't matter.  
By a semantic routine, we just mean a set of statements 
in the semantic language which is identifiable in some 
way.   Each may be a separate procedure, each may be a 
single substatement of a large case statement, or the 
semantic language may provide a special "semantic label" 
to identify the beginning of a semantic routine."

Compiler Construction for Digital Computers, David Gries, 1971, 
John Wiley and Sons, pg 167. "Calling Semantic Routines"


Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: Uche Ogbuji [mailto:uche.ogbuji@fourthought.com]
Sent: Monday, June 04, 2001 10:36 AM
To: Christian Nentwich
Cc: Jonathan Borden; James Clark; Rick Jelliffe; xml-dev@lists.xml.org
Subject: Re: Type and Structure Re: ASN.1 and XML


> > Yes, and I don't see how this contradicts James's point.  Typing belongs
> > in the semantic model and James points out that TREX and RELAX don't
make
> > the (IMO) mistake of interposing typing between the syntax and
semantics.
>
> If you think it's a mistake, blame the compiler people who came up with
> annotating syntax with semantic constructs. Personally, I think it's
> quite a convenient way (that still leaves the issue of whether XSD
> should include it unresolved though).

The compiler people rarely make a neat break between syntactic and
semantic processing.  YACC has

start   :       exp1
                { exp_value = $1; }
        ;

So the semantic constructs (implementation of behavior) is right in there
with the grammar.

I think that markup languages occupy quite a different space.  The idea is
to *separate* the data from the processing, which to me means that we can
just blindly import ideas from compiler theory any more than we can from
object-oriented development.


-- 
Uche Ogbuji                               Principal Consultant
uche.ogbuji@fourthought.com               +1 303 583 9900 x 101
Fourthought, Inc.                         http://Fourthought.com
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python