2017-04-25 14:41:48 +02:00
|
|
|
This is a mini-scheme interpreter, even though the files are all named
|
|
|
|
|
lisp*.{c,h}.
|
|
|
|
|
|
|
|
|
|
This interpreter is *extremely* slow. I can think of lots of ways to
|
|
|
|
|
improve its memory usage and performance (a factor of 5 improvement
|
|
|
|
|
seems easy), but it is very robust and it works. :) Besides, it turns
|
|
|
|
|
out that the bottleneck in most common tasks is magic itself, not the
|
|
|
|
|
interpreter.
|
|
|
|
|
|
|
|
|
|
The memory usage of this interpreter is ridiculously high. Collect garbage
|
|
|
|
|
often. :) Garbage collection is done automatically based on the
|
|
|
|
|
variable scm-gc-frequency at the top-level . . . i.e., when you see
|
|
|
|
|
magic's ">" prompt. To collect garbage at intermediate points in the
|
|
|
|
|
computation, you have to call "collect-garbage" explicitly.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
-Rajit Manohar <rajit@csl.cornell.edu>
|
|
|
|
|
Computer Systems Laboratory
|
|
|
|
|
Cornell University
|
|
|
|
|
Ithaca NY 14853
|
|
|
|
|
http://www.csl.cornell.edu/~rajit/
|
|
|
|
|
|
2020-05-23 01:10:54 +02:00
|
|
|
$Id$
|
