Mon Jun 23 18:04:02 2003
Jao, Fare and I have been corresponding about zzz over the past few days, in
an attempt to sort out what needs to be done, etc.
Jao had been working on a subset of the Tube ported to mzscheme by David
Halls. Jao further mzschemified the code, adding modules and units and a few
other features (compression, etc), as well as marking it up with
I brought up a couple of questions.
1) Why is a mobile code system needed for zzz?
What I've gotten so far is that it is actually *not* necessary, but would
allow for the most flexibility and power (and that's what we're all about here
at tunes, right?).
Jao even suggested developing the tube and zzz in parallel, and when the tube
is finished, integrating it in.
2) Why should the tube be used over other solutions such as jocaml and erlang?
Jao didn't have much to input here, but Fare suggested an orthogonally
persistent erlang with macros might be sufficient, which to me means we should
probably just stick with the tube.
We decided that mzscheme modules/units were not worthwhile, and neither was noweb.
We've decided instead to write semiportable scheme using r5rs/srfis. mzscheme
will remain the primary implementation, and we are switching over to
http://cliki.tunes.org/Elucidative%20Programming, as an experiment. This is
planned as a prototype only, with the possibility of rewriting in CL (using
That's the gist of the correspondence; I encourage Fare and Jao to correct me
if I misinterpreted any points.
I'm asking Fare and others in the know about zzz to write up a few more
substantial examples of what might be done with zzz. I'm thinking code here.
It should evolve as the prototype evolves, and eventually be test code.
Also note that fare has been busy on cto, clarifying some points. You can see
them here: http://cliki.tunes.org/zzz
http://cliki.tunes.org/Tube and as always http://cliki.tunes.org/Recent%20Changes