Re: Reuse of Semantic Specifications

  vaidyana@willis.cis.uab.edu (Viswanathan Vaidyana) writes:
> I am working on a project of developing a framework for Reusing Semantic
>Specifications of one programming language in order to define that of
>another. I am developing an object-oriented framework (or semantic
>workbench, if you may) from which it will be easy to design the semantics
>of different languages.
>
> I am interested in knowing about any other work in this area
>of reusing specifications of languages, reusing compiler components, etc.




I'm working on something of a dramatically reduced subset of what
you describe, with a slightly different slant. In my case I'm trying
to create a re-useable framework for specifying the semantics of
meta-interpreters for a particular interpreted language, Maple.
By "meta-interpreter" I mean something like what Shapiro did in
his construction of a debugger in Prolog for Prolog, described in
his thesis and in his book "The Art of Prolog".


In my case, the re-usability I'm focusing on is in generalizing the
mechanics of applying different semantic models to a fixed family of
abstract syntax trees, which basically boils down to managing how changes
to instances of the model (eg: bindings in environments, allocation of
storage, etc.) are maintained and passed as arguments to the appropriate
semantic mappings while traversing a tree. The paper is in progress
(slowly), and hopefully will be delivered at the next MSWS (Maple
Summer Workshop and Symposium). For some background on where I started
from you can read my paper in the MSWS '94 proceedings published by
Birkhauser:


Reid M. Pinchback, "Denotational Semantics Applied to the
                Typesetting of Maple Expressions", pg 22-27,
Maple V: Mathematics and its Application
Proceedings of the Maple Summer Workshop and Symposium
Robert J. Lopez, Editor


The paper is at a pretty simple level, it just assumes that you have
a very minimal understanding of denotational semantics and some
familiarity with Maple. The first few chapters of the Watt book
that Peter Mosses mentioned in his posting is more than sufficient
background for the level of DS knowledge assumed in my paper.




Re-usability of the components of the semantic models themselves
would be obviously be useful complement to what I'm working on now.
It is not part of my current effort, although I've done some
groundwork on implementing a constructive approach to category
theory (a la one of the Prentice Hall books "Constructive Category
Theory", can't remember the authors name at the moment) so that later
on I can hopefully achieve some algebraically robust flexibility and
generality in how the semantic models are constructed.






--
===============================================================
= Reid M. Pinchback =
= Senior Faculty Liaison =
= Academic Computing Services, MIT =
= =
= Email: reidmp@mit.edu =
--

Return to the comp.compilers page.
Search the comp.compilers archives again.