Re: incremental compilation via shared library

On Mon, 30 Aug 1993 16:28:18 GMT, Brent Benson
(brent@jade.ssd.csd.harris.com) wrote:


    I'm not sure how shared libraries come into play, but this sort of
    incremental compilation has been offered in Lisp programming
    environments for years:


    > (defun square (n) (* n n))
    SQUARE
    > (compile 'square)
    SQUARE
    > (disassemble #'square)
    [ ... ]


    While your proposal sounds interesting, I think the biggest obstacle
    to incremental compilation in traditional environments (C/C++) is a
    dynamic object file loader.


Well, you can hack one on most any BSD style environment -- indeed the
'-A' option to 'ld' and successive equivalent variants was introduced
to support extending Franz Lisp. SVR4 of course has a built-in dynamic
object file loader.


    [ ... ] These things are available, along with a threads mechanism,
    in the Xerox Portable Common Runtime (PCR). Unfortunately, it is not
    ported to any system that I regularly use.


An interersting data point: I once saw a very, very nice Objective C
based system called RMG developed in a joint HP-Stanford project. It
was very Smalltalk-like, but in color, and very much faster (everything was
compiled and dynamically loaded, class by class, or method by method).


Too bad I don't know what has become of RMG, which was perhaps about as good
or better than Smalltalk or NeXTStep. Does anybody at HP or Stanford know?
--

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