Fwd: Re: Maude licensed under GPL, 2.0 approaching

[Massimo Dentico]

His replay.

------------------- Forwarded message ------------------------------------
From: Brian T Rice <water@tunes.org>
To: Massimo Dentico <m.dentico@virgilio.it>
Subject: Re: Maude licensed under GPL, 2.0 approaching
Date: Sun, 1 Jun 2003 17:42:29 -0700 (PDT)


On Mon, 2 Jun 2003, Massimo Dentico wrote:

> I'm curious: there is a private mailing list for developers only?

It's an informal list for the alpha releases.

> [..]

> My idea for a next step for Tunes is to advance in parallel different
> aspects of the system:
>
> 1) to explore what we can "steal" from Maude as high level declarative
> language and what to change or inject to approach Tunes ideas;

Right. This means stealing the rewrite engine and learning how the parser
works and turning that into something nice for Tunes.

I think both of these need some work, though. But they're great as is.

> 2) how to express Maude implementation in Maude itself or how to bootstrap
> our Maude/Tunes (M/T from now), or:
>
> a) how to express in M/T the transformation from term rewriting
> to BDD (I don't know if this makes sense: I'm speaking without
> a true knowledge of the Maude implementation). In case for M/T
> we can research how to go beyond term (or graph/string/whatever)
> rewriting and capture the concept of rewriting in general,
> abstracting from the nature of base objects with our abstract
> notion of object (or someone have done this already?);

Interesting. I don't know if this has been done. It might be a big "ball
of wax" to deal with since you are talking about object identity and
(often enough) versioning. And you need some criteria to pick one
formalism over others.

I also have a suspicion that full generality in one framework is not
possible, or that the whole issue is somehow self-defining. It's much like
Bawden's description of state vs. statelessness in his Linear Naming phd
thesis.

> b) how to model low level (machine) entities, beginning from the CPU,
> in M/T;

That's probably not difficult.

> c) at last, how to map the transformation in a) in terms of entities
> in b);

Rewriting? :) Seriously, this is something that I've been thinking about
as a meta-meta-level module interface to alter the module's rewriting
strategy to effectively alter the course of the implementation.

This is best understood in the knowledge that ultimately, rewritings to
represent isomorphisms will be bidirectional, and there will be several of
them to choose among. (Think of the Sprint results.)

Now, this is also difficult in that we'll have to develop a whole
vocabulary for it, and that these meta-strategies would have to be cross-
modular.

> 3) co-evolving the spec with feedback from 1) and 2).

Right.

> Note that in a) the underlying assumption is that the implementation of
> Maude is state of the art wrt rewriting systems but I'm absolutely
> conscious that, for the nature of the system itself, this choice is not
> mandatory.

Sure.

> I remember also a list of "experiments" (in and about Tunes) in a
> e-mail by Faré on the LL1 mailing list but I have not a reference
> at hand. IIRC, some are about self-organization, emergent properties
> and alike. On the subject I don't know if you have noticed that parts
> of Maude are to be used in ADATE's successor, see bottom of page of:
>
> http://www-ia.hiof.no/~rolando/adate_intro.html

Interesting. I think I was aware of this some time ago, but forgot.

> Of course we need to show how better, wrt other usual approaches, all this
> stuff actually solves real problems.

I really want to forward this to the list, but I'm really wondering what
good it will do? Forward (or bounce it to avoid the extra level of
indentation) it yourself if you like.

-- 
Brian T. Rice
LOGOS Research and Development
http://tunes.org/~water/

------------------- End of forwarded message ----------------------------

Regards.

--
Massimo Dentico