Re: Project on CMU-LISP

[Rob_MacLachlan@LISP-PMAX2.SLISP.CS.CMU.EDU]

    Date: Fri, 14 Feb 92 11:33:41 -0500
    From: Chris Schmechel <schmech@cs.unca.edu>
    Message-Id: <9202141633.AA03808@ivy.cs.unca.edu>
    To: Rob_MacLachlan@LISP-PMAX1.SLISP.CS.CMU.EDU
    Subject: Project on CMU-LISP 
    
    Hi,
    
    I would be interested in porting the sources to Ultrix.

It may not be as easy as I thought.  We were considering doing the port
ourselves, so we investigated the Ultrix "mmap".  As near as we can tell
(in the kernel we have), mmap can't map files, which is what we want it
for.  It would still be possible to work by "read"ing in the whole image,
but this would make startup slow, would prevent any sharing, and would take
up lots of paging space.
 
    I would like to start some research collaboration with CMU, due to the
    fact I am not in a major-research oriented environment.  I am
    interested in the reimplementation of the foreign function interface
    and also the native CLOS.

Well, you're welcome to poke around in the code, make suggestions, ask
questions, etc.  It's hard for me to be more concrete than that right now,
since what we have clearly planned in those areas, we've mostly already
done.  There's lots left to do, but it's very vague.

Another thing you could do is work on adding some more compiler
optimizations.  For example, there is no instruction-scheduling
optimization on the SPARC.  Or you could write loop-invariant optimization
for the back-end.  Or enhance the register allocator.

    What are the possibilities of collaboration...
    
Well, if you do something interesting, you can always author or co-author a
paper.

    I also plan to apply to graduate school at CMU.

I should warn you that:
 -- The CMU CS grad program is very selective, and
 -- Work on CMU CL probably won't help in admission.  (Though published
    papers could.)

The only faculty member on the project is Scott Fahlman, and he is research
faculty (not tenure track.)  His main research interest is connectionist
AI, with Lisp being a (successful) sideline.  CMU CL has not really been a
research project in the past -- we haven't ever had a grad student.
Although I have done some researchy stuff, I'm staff, not a student.

  Rob