perm filename LETR1[DOC,AIL] blob sn#050631 filedate 1973-09-23 generic text, type C, neo UTF8
COMMENT āŠ—   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002				26 JUNE 1973
C00005 ENDMK
CāŠ—;
			26 JUNE 1973
			Artificial Intelligence Laboratory
			Dept of Computer Science
			Stanford Univ.
			Stanford Ca. 94305
			(415) 321-2300 x4971

Prof. N.F. Stewart
Universite de Montreal
Case postale 6128
Montreal 101, Canada

Dear Prof. Stewart,

	Very sorry about the delay in answering your inquiry 
about SAIL, but I've been out of town until recently.  SAIL
is implemented, as you may know, for a PDP-10, and is available
from DECUS.  Currently, we are producing a new SAIL release
to supercede the current DECUS version, which is about three years
old and lacks many features found in the newer SAIL.  This 
release will be sent to DECUS as soon as the manual is printed
(within two weeks, I hope).  As to getting SAIL to fit into
a 64K time-shared PDP-10, it would be a close thing.  The compiler
was written for a system with a more reasonable amount of memory.
I don't have the exact figures for the export compiler, but
I expect that the compiler will require 25K or more to start up.
For medium size compilations, this grows to 35-40K.  Assuming
that the DEC time sharing system takes 30K, this is getting tight
indeed.  The old SAIL -- I.e. that currently available from DECUS,
is a bit smaller, but it would still be a tight fit.
Once your code is compiled, of course, the storage pinch is
less restrictive.  Typically, a SAIL image will use anywhere from
3K to 8K of routines from the runtime library. (Here at Stanford,
we keep the runtime routines in an 8K shared upper segment).
As near as I know, SAIL has never been implemented on an IBM
machine.  We should have some more definite numbers when the
DECUS release finally goes out.

			Yours truly



			Russell H. Taylor