HLL and misa...

[Francois-Rene Rideau]

>>    Wow ! I wish I could raise that much money ! How did you do ?
> we got all of it.  we've only used a tiny bit.  we're really saving up
> to but a computer in the fall.
Well, please tell me how you could convince them. Who are they ?
What is your exact situation ? How did you meet them ? What arguments
did you use ? What constraints did they put on your help ?

> we were very surpised to get it.  our funder seems to have more faith
> in us than we have in ourselves! :)
   I hope that you won't disappoint them. I wish I can help !


>>    Hey ! If you're looking for help and it can be used as a pre-thesis
>> training period, count me in !
> i am interested in discussing the language with you.  i was very
> surprised when i read your pages because they mirrored almost exactly
> our thinking about language and OS.  so far very few people understand
> the ideas that we have.
   I have always been sure I wasn't the only one with those ideas,
because I didn't invent anything. I just wrote down ideas that were
"on the air", everywhere implicit, nowhere explicit. Just their
obviousness appeared to me before it appears to everyone else.
That's why I call TUNES a "revolutionary" project (beware the multiple
meanings of this word -- see the TUNES Glossary): the changes it
brings will completely change the world, yet, they'll seem obvious to
everyone and noone will be able to tell who saw it first.
   BTW, I'm in trouble with respect to school: if I don't want to
lose a year, I need do a training period and publish a technical
report about it before september ! I'm currently on the trail of
something in France, but if you can get me something serious in the
UK, I'll jump at once !


>>    Well, though I agree s-exps have advantages, and may be used at first,
>> I don't plan to do that in TUNES, because it kinda links too much the
>> semantics to a particular representation. The Dylan people chose not to
>> continue with s-exp anymore, and I think they're right...
> 
> please eloborate.  this was a _very_ big debate for us.  i was
> orginally on the side of developing a new syntax but eventually was
> swayed.  perhaps it is not fair to say that our syntax is
> s-expressions but is descended from them.
   What I mean is that s-exps, *as used in LISP* are too much of a
concrete representation: LISPers have definitely integrated the
fact that lists will be implemented with cons cells, and use cons
stuff (cons, car, cdr) as a low-level way to access data.
   This is definitely *BAD*. I like lisp as a high-level language.
I *hate* lisp as a low-level language.
   To me, thinks should stay as high-level as can be, and low-level
things should be specifiable independently. It should be possible to
program a high-level thing, then map it into a completely different
low-level implementation than cons cells.
   Sure, any working environment should provide a default way to implement
things, and cons cells are an easy way for that. But this really *should
not* be part of any language standard.
   For instance, I'd quite prefer an ML-like abstract syntax, but
with automatically (and programmably) generated constructors, deconstructors,
parsers, printers, and functors in general.
   Also, something *very bad* about such low-level representation as the
LISP stuff, is that it makes dynamic annotations very hard and dirty.
E.g. in lisp, being able to parse (interactive ""), type declarations,
and such, make things *very* dirty.


> it is very useful in AI (which i study) to be able to modify your code
> during runtime.
   I fully agree. But why only in AI ? In everything !
BTW, there is something about AI that I did not write in the TUNES papers
yet, but will add the the WhyNewOS article ASAP:
* intelligence is the fruit of a long tradition.
Even a most precocious human being must be bred for years before yielding
the faintest result, not talking about the time meant for developping
a fine breeding environment (e.g. our dumbest ancestors were almost
identical genetically; we spent tens of thousands of years developping
our existing culture and education).
* computers currently do not allow any information to persist reliably
more than a few months, and won't translate information from old software
to newer ones. In no way of reasonable cost can data cooked by the computer
persist over changes in a program.
* Hence, intelligence is not possible with existing architecture.


> s-exp are currently the best code structure for
> dynamic modification of code.
   Yeah: *currently*.
I'm for allowing anykind of representation, but automatically
generating functors.

> i am not saying that it is ideal but i
> think that it would be useful to develop a defined structure (which of
> course would be redefinable like everything else, and be an object,
> ...) so as not to inhibit this runtime modification.  i know of no
> other syntax which gives you this power like s-exp.
   s-exps are a good thing *to begin with* do not make the same mistake
as lisp. Do not specify a one low-level representation for high-level
constructors.


> perhaps it is time to catch up on my french (i studied for 4 years and
> then russion for two - it has been a long time).
   Hehe. I'd be glad we communicate in French. However, people on the
TUNES list come from all over the world, and I fear English is the only
fit language :( :( :( :( :(


> are you studing in france?  where do you live?
   I am studying Computer Science in Paris France, at the ENS
(Ecole Normale Superieure) of rue d'Ulm (if that tells you anything).
I also happen to live in Paris (lucky I), which is a very beautiful town,
though pollution is making it less bearable every year (but we're far from
what happens in Athens, happily).
   If ever you come to Paris, please come by ! My parents also have a house
in Britanny. If you ever come, please pass by, too !
[Note: this also stands for everyone on the list: if any of you come to
France, I'll be happy to welcome him at home !]

--    ,        	                                ,           _ v    ~  ^  --
-- Fare -- rideau@clipper.ens.fr -- Francois-Rene Rideau -- +)ang-Vu Ban --
--                                      '                   / .          --
Join the TUNES project for a computing system based on computing freedom !
		   TUNES is a Useful, Not Expedient System
WWW page at URL: "http://www.eleves.ens.fr:8080/home/rideau/Tunes/"