Laurent Martelli martelli@iie.cnam.fr
02 Jan 1999 00:33:53 +0100

>>>>> "Brian" == RE01 Rice Brian T EM2 <BRice@vinson.navy.mil> writes:

    Brian> for instance, can anyone PRECISELY explain why Self, Lisp,
    Brian> Scheme, ML, Modula, Smalltalk, Oberon, or (insert favorite
    Brian> computing interface here) is not Tunes?  Why is it that you
    Brian> can't build Tunes easily within these systems?  Because
    Brian> they do not transcend language, mathematics,
    Brian> finite-structure models, ...  You need a language that can
    Brian> concisely describe an entire hardware system and its
    Brian> operation, as well as consolidate redundant information
    Brian> (i.e. reduce information noise in the general sense) in a
    Brian> systematic way.

I only know Lisp and Smalltalk. And I think thse languages would be
almost -- if not completly -- suitable to implement the computer
system of my dreams. Because they provide high level abstractions to
their users. They are both based on _one_ simple concept : function
call for Lisp, message sending for Smalltalk. However, they both lack
a more radical separation of interface and implementation. I think
that we should be able to define services first, that is specify what
a service does. And then provide one or more implementation of the
service in case the system is not able to build one from the
specs. And there should be a way to tell which implementation of
service you want to use.