[Python-logic] (no subject)

Nicolas Chauvat Nicolas.Chauvat@logilab.fr
Thu, 16 May 2002 13:02:32 +0200 (CEST)

> (Note: When I use the second person pronoun "you" herein, I refer
> specifically to Nicolas Chauvat rather than to the list as a whole.)

Hey it's me :-)

> I'll admit that I'm interested in the prospect of logic programing for
> Python. However, I'm a little confused and doubtful about the status
> of this "unofficial" SIG. First off, could you please clarify the
> founding history of the SIG? On the Meta-SIG there was the notice of
> intention to create the SIG, one positive reply, and then a message by
> Nicholas Chauvat stating that SIG creation had been "refused," but no
> "official" statement of SIG rejection by Guido et al. Was official
> refusal given in a private e-mail, or did you withdraw the proposal by
> yourself?

I had a private chat with Guido van Rossum and Andrew Kuchling (IIRC).

> I'm also concerned by the use of the word "refused" - to me (and I may
> be wrong) the word has connotations that there was active opposition.
> If there was active opposition, I wonder who was opposed, as I could
> see no negative comments on either the Meta-SIG or on c.l.py.

The opposition was stated as follows "you only got one answer to your
query for a SIG creation, and the fact that it was a positive one and that
you got 10 positive answers on c.l.py is not enough to prove that this is
a viable SIG. Run it by yourself if you wish, it will be brought back into
Python.org if it happens to work out".

> Regardless, I'm also curious as to what the goal of the SIG is.
> Addition of a new computational paradigm to Python's arsenal is a
> somewhat major step, and would require favorable winds from the BDFL
> to avoid the logic-SIG from turning into another lambda. You mentioned
> in the Logic-SIG creation letter that you had been in contact with
> Guido. Could you share what he had to say, so people can get a better
> impression of the framework the SIG works under?

At first I e-mailed Guido van Rossum (you know him), Christian Tismer
(The Stackless Guy), Denys Duchier (one of the Mozart/Oz guys) and 
Alastair Burt (DFKI.de, who I chatted with about CLP+Python) offering to
organize a Bird of a Feather meeting about logic programming during
EuroPython. Guido told me not too wait for that long and get the thing
started right away.

Why these four people ? For Guido it's obvious. For Alastair, because we
chatted about it and he gave me several ideas to explore. For Christian,
because Alastair told me it could be interesting to code propagators (read
the mozart manual) as continuations or microthreads. For Denys, because he
wrote part of Mozart, is a python fan, knows more about logic-programming
than I ever will and could have provided very useful implementation ideas
and guidelines. Unfortunately, it appears he is not interested.

> As I see it, there are several levels at which it could operate:
> A) Guido loves the concept - do what you have to.

If I understood correctly, Guido has had very few to no experience with
logical programming and first wondered why I would want that kind of
features in Python and what they would be useful for.

My (very) short answer to his question : i'd rather stick to python and
its huge user base and library even if I have to spend time working on
adding new features to the language than move to mozart whose syntax is
for me too awful to outweight its greatness (if it wasn't for the syntax,
it is THE language I would be using, just read the list of its features
from the web page http://www.mozart-oz.org).

> B) Guido likes it - it's implemented, but no new keywords -
>         new operators (like ?+, as a random example) are OK.
> C) Guido likes it - but not at the cost of syntactic change - 
>         internal changes (as in the True/False addition) are OK, though.
> D) Guido is fine with it - but only as module in the standard library.
> E) Guido doesn't care - it's implemented like Numeric - a quasi-official
>         package distributed separately from the main distribution.
> F) Guido hates it - it is an outcast package, living in the shadows,
>         begging for food in the street gutters ...

I think it's more like "I've no use for that but understand someone may
want it. Prove it works and people like it, then you'll probably get

Now, the way *I* would like it is more like option D, and C if we really
cannot do otherwise (but at this point I doubt we'll need C).

> >From the SIG description posted on the Meta-SIG mailing list it appears you're leaning toward options D or E:
> ------------
> Logic SIG: Exploring logic and constraint programming with Python
> Description: This SIG is for discussing and comparing several
> approaches to implement logic and constraint programming
> techniques in Python.
> Deliverables: A set of modules for people to easily describe and solve
> rule-based problems and constraint-based problems in a pythonic way.
> -------------
> Sorry to make this post seem like an interrogation.

I'm the kind of people who very much like to get interrogations. It makes
me get my thoughts in good order and sort strong and weaks justifications.
It makes other people clear about the goals and ideas, then they can build
on it.

> It's just that I'll bet, if my personal feelings are any indication,
> that more people would be willing to participate if the uncertainty of
> an "unofficial SIG" was cleared up, and if they were given a realistic
> view of what is hoped to be accomplished.

If ten people were now to post to meta-sig asking for the logic-SIG to
become "official" right away, I'm sure it would. Otherwise, Logilab will 
continue its works to provide logic programming features to Python, in
the same way Christian Tismer does for stackless.

> Rocco Moretti

Welcome ot our "semi-official" Logic-SIG, Rocco :-)

Nicolas Chauvat

http://www.logilab.com - "Mais oł est donc Ornicar ?" - LOGILAB, Paris (France)