[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: a (revised) draft of R5RS



   From: Jeff Dalton <jeff@aiai.ed.ac.uk>
   Cc: rrrs-authors@martigny.ai.mit.edu
   Date: Thu, 20 Mar 97 02:55:34 GMT

...

   In any case, I think the only answer for those who are not satisfied
   with Scheme more or less as it is now is for them to develop a new
   language.

   -- jeff

Okay.  Let's design a language modeled after Scheme without the
constraint of being compatible with Scheme.  I suggest that the focus
be on supporting multithreaded and distributed applications.  To
support multithreading, I suggest we make some common data structures
immutable.  Pairs are the most obvious candidate.  I also suggest that
lambda bound variables and letrec bound variables become immutable and
we restrict the kinds of expressions that can be used to initialize
letrec variables to lambda expressions.  In addition the results of
EQV? applied to two procedures should be undefined.

Next we need a threads package that makes independently scheduled
threads available to applications.  And a distributed programming
system not too much unlike the work from NEC.  And .... oh, who am I
kidding?  Java has already filled this void.  Maybe Scheme should just
remain as an elegant jewel for all to look at.

John