Hey, in all my browsing of Lisp history, I don't recall mention of
implementation of Lisp on Burroughs mainframe architecture (I think it
is primarily instantiated now as a virtual machine atop Intel hardware
at Unisys).
It is heavily stack oriented and has tagged memory. Tags indicate
machine-aware data types, not user data types, but I was wondering if
there is any benifit of this architecture for Lisp implementation.
For instance, w/regard to the segment based virtual memory scheme?
Did anybody do it?
Maybe even there are some happy Lispers out there currently Lisping
away on 8800?
Daniel,
a 1975 master's thesis "Lisp 1.5 and an implementation in the system
B-6700" by this guy was the very first graduate level text I read (I
actually bought a copy of it):
http://www.ime.usp.br/~song/index2.html
I don't think he did a good job as he worried too much about m-
expressions and not enough about s-expressions (well, math guys do
love their little greek letters).
A few years later I used Reduce on that same machine and it was a
great Lisp even if it did have an Algol syntax. I would be surprised
if no traditional Lisps were developed for this architecture.
-- Jecel
On Jul 16, 10:13 pm, Jecel <·····@merlintec.com> wrote:
> a 1975 master's thesis "Lisp 1.5 and an implementation in the system
> B-6700" by this guy was the very first graduate level text I read (I
> actually bought a copy of it):
Actually the B-6700 is the first mainframe machine I ever touched by
hand (it was in 1975, too) and the 6700 remains the primary driver of
my curiousity about the whole architecture.
Thanks for the link!
A search on Reduce turned up a paper in the ACM library for the
B-1700, and the authors claim to have implemented a Lisp s-exp
compiler on that architecture to support Reduce. Therefore, it might
be that there was a fairly true Lisp under Reduce on the 6700 - I
wonder.