ganesh@sbcs.UUCP (Ganesh Gopalakrishnan) (03/05/85)
I would like to invite the readers comments (and concrete information) on the idea of a CLU to Lisp compiler. I see the following as strong arguments for having one. The arguments are from the point of view of a person who REQUIRES Lisp code, ultimately: (a) CLU provides a rich set of abstraction mechanisms and compile-time checks. Often it is the ideal language for programming. (e.g. if one were to write abstract data-type definitions, CLU is superior to Lisp, in my opinion). Most such checks are absent in a Lisp environment. Also, it is hard to guarantee/enforce the concealment of the data representations, in Lisp. (b) Due to the semantic closeness of Lisp and CLU, (both can support object oriented programming, both allow infinite and recursive data-structures, etc.), writing such a compiler does not seem to be hard. It ought to be possible to make the generated Lisp code very readable and understandable; (the relationships between the CLU source and Lisp target must be preserved as far as possible.) This opens up the attractive possibility of debugging in the Lisp environment, too. (Interpretation as opposed to the often slow Compile->Execute cycle). (c) Possible to interface/mix systems written in CLU and in Lisp (e.g. Reve theorem prover[CLU] and RRL theorem prover[Lisp]). Is there such a compiler in the public domain ? Many thanks, in advance.