Re: Problems with Hardware, Languages, and Compilers

Herman Rubin (hrubin@stat.purdue.edu) wrote:
> what is needed is that other operations can be added at will to the
> language, and that the compiler can be instructed to select from
> optimization techniques which the user can supply it, as well as
> from those which the compiler writer included.


You might want to look at the "meta-programming" techniques which have
been developed in the context of Lisp (for details ask comp.lang.lisp).


Using compile-time transformations, called macros (not to be confused
with the primitive stuff of the C preprocessor and many assembler
macro facilities), to define complicated optimizations and new
abstractions is an important part of the Lisp culture (Henry Baker
once called Lisp macros the ultimate CASE tool, and agree with that).


I remember having read about some work at Xerox PARC with protocols to
influence and extend the decisions of a compiler about representations,
and Paul Wilson has written a lot about the merits of powerful macros
which can interact with the compiler's knowledge about the program.


Even if current implementations may perhaps not be able to deliver what
you need (I just don't know), I think the concepts are appropriate to
facilitate the kind of extensibility in which you appear to be interested,
and certainly worth knowing in case you want to create something "better"
(for your requirements).


-- Marc Wachowitz <mw@ipx2.rz.uni-mannheim.de>
--

Return to the comp.compilers page.
Search the comp.compilers archives again.