Work

[Faré]

If some of you are still interested in working on TUNESy ideas, here
are some general projects that you may want to work on. More
suggestions welcome, and/or pointers to things that already do most of
the work.

* Work on Slate, give it a decent virtual machine implementation
* Start from Ian Piumarta's COLAs (or the same idea), and build some
interesting language implementation for, e.g. Slate, or a good Lisp
system.
* Start from Movitz, build better system abstractions on it or for it.
* Start from Exscribe, show how a high-level language can benefit a
domain where the only programmable system (TeX) is a complete mess.
* Start from Philip-Jose, the "farmer" I just wrote, and extend it
with fork() and a better communication protocol, so as to get
Erlang-like concurrency in Lisp
* Starting from some Lisp compiler, implement the "user-declared
safe-point" protocol from my thesis, and build a reflective
implementation framework with reversible compilation, etc.
* Starting from some expert system, build some expertises for
analyzing and refactoring programs, e.g. abstract interpretation,
partial evaluation, deforestation, aspect-oriented programming, logic
metaprogramming, etc.
* Build a library (for Lisp or Slate) combining a monotonic fragment
and a linear logic fragment, and build user-visible abstractions from
that.

If we build up a bit, we may be in a situation that we can apply for
GSoC sponsorship next year. We missed 3 such events. Mea Culpa.

[ François-René ÐVB Rideau | Reflection&Cybernethics | http://fare.tunes.org ]
Wishing without work is like fishing without bait.
-- Frank Tyger

[Kyle Lahnakoski]

Faré:

Do you have a link to your "user-declared safe-point" thesis?

Do you already have a plan for reversible compilation? How do you
prevent the loss of high level intent during compilation? Do you plan
to annotate the low-level code? Do you contend that the act of
compilation does not loose (useful) information? Or, maybe something
in between?

Thanks

Post by Faré
If some of you are still interested in working on TUNESy ideas, here
are some general projects that you may want to work on. More
suggestions welcome, and/or pointers to things that already do most of
the work.
* Work on Slate, give it a decent virtual machine implementation
* Start from Ian Piumarta's COLAs (or the same idea), and build some
interesting language implementation for, e.g. Slate, or a good Lisp
system.
* Start from Movitz, build better system abstractions on it or for it.
* Start from Exscribe, show how a high-level language can benefit a
domain where the only programmable system (TeX) is a complete mess.
* Start from Philip-Jose, the "farmer" I just wrote, and extend it
with fork() and a better communication protocol, so as to get
Erlang-like concurrency in Lisp
* Starting from some Lisp compiler, implement the "user-declared
safe-point" protocol from my thesis, and build a reflective
implementation framework with reversible compilation, etc.
* Starting from some expert system, build some expertises for
analyzing and refactoring programs, e.g. abstract interpretation,
partial evaluation, deforestation, aspect-oriented programming, logic
metaprogramming, etc.
* Build a library (for Lisp or Slate) combining a monotonic fragment
and a linear logic fragment, and build user-visible abstractions from
that.
If we build up a bit, we may be in a situation that we can apply for
GSoC sponsorship next year. We missed 3 such events. Mea Culpa.
[ François-René ÐVB Rideau | Reflection&Cybernethics |
http://fare.tunes.org ]
Wishing without work is like fishing without bait.
-- Frank Tyger

[M. Edward (Ed) Borasky]

Post by Faré
If some of you are still interested in working on TUNESy ideas, here
are some general projects that you may want to work on. More
suggestions welcome, and/or pointers to things that already do most of
the work.
[snip]
* Start from Philip-Jose, the "farmer" I just wrote, and extend it
with fork() and a better communication protocol, so as to get
Erlang-like concurrency in Lisp

Gambit Scheme already has this -- do a Google search for "termite".

[Faré]

Post by M. Edward (Ed) Borasky

Post by Faré
* Start from Philip-Jose, the "farmer" I just wrote, and extend it
with fork() and a better communication protocol, so as to get
Erlang-like concurrency in Lisp

Gambit Scheme already has this -- do a Google search for "termite".

Yes -- I but needed it for practical purposes, wherein it had
* to be written in Common Lisp, to integrate into existing
applications, reuse some libraries, be portable, benefit from its
compilers, etc.
* to support robustness at the process level (workers WILL build
cruft, bug out, deadlock and otherwise have to be kill -9'ed)
* to be able to spawn remote nodes through ssh
Philip-Jose is pretty minimal, inefficient and under-featured, as far
as distributed system infrastructures go. It was designed to do
repeatedly drive one big job I have at work. As a result, it is not a
generic solution, though it purports to be extensible into one -- but
it does work.

That said, if I had to extend Philip-Jose some more, I would
definitely have it reuse the interfaces of Termite.


My general message for TUNESers is that for the survival and growth of
the memes we care for, we have to care about expediency. There is no
long term without short term. The evolutionary path matters if we are
to reach a goal. If humanity is worth saving, it is worth saving at a
profit. Insert your favorite another trite saying.

One of the many sins of my youth was to not take this seriously
enough. I hope I'm learning.

[ François-René ÐVB Rideau | Reflection&Cybernethics | http://fare.tunes.org ]
Don't tell me how hard you work. Tell me how much you get done.
-- James J. Ling