Interlisp versus MacLisp

[MILLER at MIT-AI (Mark L. Miller)]

CC: KVL at MIT-AI, MILLER at MIT-AI, Jonl at MIT-MC

    Since RMS seems to have opened this can of worms again, let me throw in
a new remark.  Having just left the I.T.S. world for the Twenex world, I
have a greater understanding of why Twenex sites tend to prefer Interlisp.
Simple: Interlisp provides nice support for things Twenex users need to do,
such as (SYSTAT), ^T, named lower forks, mail maintenance, reminders, JSYS
calls, etc etc.  MacLisp leaves the burden on the user to try to figure
out, e.g., how to implement a JSYS or (TENEX -- ) etc. function.  I may try
to find time to build in such things (various people have been helpful in
providing suggestions).  Or, I could just switch to Interlisp.  (When we get
more horsepower here, I might.)  Of course, there is also the small matter of
documentation.
    Personally, I still prefer MacLisp.  But I doubt that MacLisp can
survive all these problems plus the proliferation of flavors (NIL, LISPM,
etc).  If Interlisp appears on the Vax, or if a 30-bit 2060 IInterlisp
appears, or if it appears on some as-yet-to-announced Xerox or BBN personal
something, I think it will win out.  As horsepower gets cheaper, those
only-partially-intelligent DWIM-like things will seem more worthwhile.  A
good example are the run time error checks in the LISPM.  
                          Regards, Mark