[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Axiom-developer] Back again.

From: Stephen Wilson
Subject: Re: [Axiom-developer] Back again.
Date: 24 May 2007 23:27:19 -0400
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.4

C Y <address@hidden> writes:

> --- Stephen Wilson <address@hidden> wrote:
> > Greetings All,
> > 
> > I participated in this community for a while in 2004-2005.  Simply
> > put, I became discouraged due to my own lack of decisiveness, and in
> > part due to a lack of `direction' I sensed in the community as a
> > whole.  I have followed this list virtually every day since that
> > time.
> > 
> > I am compelled to rejoin and contribute once again.
> Welcome back :-).

Thank-you, Cliff.

> > I realize that the goal of a Lisp translation runs contrary to many
> > opinions on this list.  My hope is that having more Boot code
> > documented will be seen as beneficial.  As for the Lisp rewrite, I
> > would love to hear from anyone who shares an interest in this
> > direction.
> I also share that interest, although not the needed expertise (as yet,
> anyway - that's what learning is all about :-)
> I have two concentrations at this stage that look achievable - one is
> to provide the primary web functionality in Lisp (this one is fairly
> close to working, and can handle at least basic files) and teach the
> Lisp asdf system to load pamphlet files directly using the lisp based
> web package.

I have looked at your work and must say I am impressed and
optimistic.  The potential benefits are enormous.

> Once those components are in working order, I hope we can begin to
> provide logical asdf packages of components of Axiom, and load them
> like normal lisp libraries.

I see ASDF as the right target.  The problem is getting the system
into a state which is `Lispy' enough for the full benefits to be

> At the moment, I'm focusing on asdf.lisp itself - converting it into a
> pamphlet file looks like the best way to understand it well enough to
> map pamphlet ideas to asdf ideas.

Oh yes.  Absolutely.  A large portion of my time will be devoted to
documenting the existing Axiom system for the sole purpose of
improving my understanding.

> I know less about the mechanics of converting boot to lisp and what
> good ideas would be, although I would be curious about using existing
> work in Lisp to make life easier. 

Existing libraries are always an option.  Unfortunately (or
fortunately, depending on your perspective :), there is a lot of code
we must write.  My main focus is on the core of Axiom.  My hope is to
engineer a system which is extensible, and thus amenable to the
incorporation of well established libraries.

> Making linedit work on the targeted Lisp platforms might let us
> avoid the need for clef, for example.

Yep. The new axiom.el might fulfill that role too for a large number
of users.

> The SPAD compiler itself I would like to see at least separated out,
> although I suspect any substantial improvements to it would be a major
> task. 

Yes, it is a major task.  But not impossible.  As I mentioned before,
I see a two stage process. A simple Boot->Lisp which yields as
byproduct documentation for both representations.  Then a Lisp->Lisp
which is much more involved and requires a global view of the system
as it stands (aquired via step 1), and a global view of where we would
like it to go.  For example, it would be in this latter stage that the
mechanics be implemented to support Aldor-style dependent types.

> I am very much interested in Gaby's work on the definition of the
> SPAD language and what can be done using that as a foundation.

I am too.  I have spent a lot of time privately thinking about SPAD
and Aldor semantics over the past year or two and do not doubt that it
is a hard problem.  Gaby's results will no doubt be of great value.

> Perhaps work like META could be used to
> advantage, although I don't know enough about this type of
> programming to properly evaluate it.

I am partial to hand written parsers myself, but am not in a position
to evaluate META.

> Anyway, this is a direction I at least find interesting.

I am glad to hear this!

> Cheers,
> CY


reply via email to

[Prev in Thread] Current Thread [Next in Thread]