ZIRUS & Tunes

Francois-Rene Rideau rideau@nef.ens.fr
Sat, 4 Jan 1997 04:39:17 +0100 (MET)


Dear Nickolay,

>   we have read your home pages about TUNES project. Your idea is beautiful.
Thanks a lot!

>   For long time we have being developed
> an Integrated Software Environment (ZIRUS)
> as a new platform for software development and implementation.

>   We consider our ideas are closed to your project.
The subject of your message was about collaboration.
I'd be glad to collaborate in any appropriate way.
Maybe ZIRUS can serve as a basis for Tunes development.
Is any source available for some basic ZIRUS software?

>   We send you our draft about ZIRUS.
I read it. Seems interesting indeed. Send me more about it!

>    We'd like to refer to your publications about TUNES in our papers.
>   Could you e-mail us about your basic publications and more closed
> another
>   projects.
Feel free to refer to what I published!
The problem is that what is on the WWW is all that is published :( :( :(
But if you'd like me to expand specifically on some subject,
tell me so I write an article about it:
it'd be very kind of you to guide me
in what parts of the Tunes project need be explained better...


Merry Christmas and a Happy New Year!


>       The conception of ZIRUS is:

>       1. Representation of all system elements on logical level
>  that is the base for external (for users) and internal (for
>  a hardware platform)  projections  of these  elements.
Ahem. IANSWYM (I am not sure what you mean).

>       2. [SEGMENTS]
>	3. The objects  are represented as a set of SEGMETs describing
> both data structures and methods.
Yeah, this is a good idea, and that's also how the lower-level modules
of Tunes should do it; ZIRUS might be an excellent start for the
lower-level parts of Tunes.
   However, higher-level modules would be more abstract,
allowing reflectively defined representations to be used.
A same logical high-level object would not need be represented by
a segment; it could be represented by a linked list of segments;
by stuff outside of the computer RAM
(accessible by disk, network, or any other I/O), etc.
This should really stay abstract.
   I can imagine that the system could automatically determine
and dynamically modify the low-level representations for a
same high-level object, so as to dynamically optimize
various phases of computations done on the object.
Actually, such techniques have been successfully used before,
though not on the scale I dream for them...

>       4. Logic of program moduls are described by objects of
> specific  structure with translation and implementation methods.
>
Do you mean your system is reflective w/ a standard low-level
representation?

>        5. Including in knowledge base logic-functional links
> allows to influence upon a process of program creation and implementation.
Do you mean you have a semi-automatic meta-programming system?

>        6. Evolutionary interrelated development of the conception,
>  software environment and applications by a mobile team.
Do you mean you have a dynamically reflective system?

>     To the end of making the ZIRUS environment internally full
> a language OLFIS has been developed and included as  the  next
> tecnologies:
What does "internally full" mean?
Perhaps standalone reflective?

>     - functional programming (programming at  the  expence  of
> introducing and decoding definitions at different conditions);
I know what I learnt as being functional programming,
but IANSWYM with your comment...

>     - description of logic in the form of functionally-bounded
> modules (structural elements of the language);
What is functionally-bounded?

>     - possibility of 1 :  n structuring transformations:  from
> one  external  module's  representation  to  a set of internal
> representations, and vice versa, from one internal to a set of
> different external representations;
Like my really abstract objects, ain't it?
Or perhaps you allow only use of representations
that are founded wrt low-level memory chunks...

>     - simulation  of  both  external   and   internal   module
> representations in the form of diverse-type segs;
Ahem. Another kind of reflectivity again?

>     - adjustability of language syntax with the aid of special
> OLFIS dictionaries;
Good. Just how adjustable is this.
How does it compare to Scheme macros,
or to Coq grammars?

>     - module's  internal  code  includes  different   elements
> defining   the   internal   logic  of  the  module  (lists  of
> parameters,  sequences,  constants,  functional and seg terms,
> etc.) and contains information links with the other modules to
> be retrieved and the data structures.
You mean accessible as first-class data?

>     - possibility  of constructing the event-type control with
> the use of processing the modified data;
I don't understand.

>        PROCESSING LOGIC-FUNCTIONAL LINKS (LFLs)
>        AND BASIC SCHEME OF PARALLEL PROCESSING
Ahem. What is an logic-functional link?
Can you give a few simple examples?
   You say they can be selected.
Is it statically or dynamically?
>From a statically determined list,
or a dynamically modifiable list?
   What is a fail LFL?
What when such a failure would make dependent computations
to yield incorrect results?
   You say A and E quantifiers can be used.
Can arbitrary mathematical/logical statements be expressed?
Doesn't that make it undecidable to check the specs?

>     Existential and universal  quantifiers  are  specified  as
> operators in the external structure of modules and may be used
> for describing complex LFLs.
Great.
But how are they checked?
Are there/will there be reflective proof tactics?

>     OLFIS allows one to describe the logic of execution, which
> may be used for parallelizing large programs and represents  a
> set  of independent processes (main modules of OLFIS stored in
> the module knowledge base). [...]
I can't seem to understand what kind of parallel computation ZIRUS models.
How does it compare to say pi-calculus,
or any other well known (or traditional) parallel computation system?

PS: I fear your english is not always perfectly clear,
and sometimes I'm not sure what you mean,
though what I guess looks promising...
Surely I do sometimes suffer from similar problems :( :( :(

== Fare' -- rideau@ens.fr -- Franc,ois-Rene' Rideau -- DDa(.ng-Vu~ Ba^n ==
Join the TUNES project for a computing system based on computing freedom !
                TUNES is a Useful, Not Expedient System
URL: "http://www.eleves.ens.fr:8080/home/rideau/Tunes/"