🚀 go-pugleaf

Hi out there.  I'm currently working on a large project written in
Turbor Prolog, Version 2.0 (Borland).  This project makes extensive uses
of databases, both global and local, being split into about 9 modules.

I have found that IMHO the internal memory management does not appear to be
working - I use a local database for some function, retract all of the
predicates within in, and then find out that the memory does not appear
to be being freed (according to the predefined storage predicate).  The
funny thing is that sometimes when I try using a different database,
when I'm finished with this the memory used by the previous database is
then freed.  This also appears to be random - I have used the same
sequence of key presses 5 times in a row once, and except for the
initial free memory at the start of the program, on every occasion the
free memory was different, according to storage.

Does anyone know
	(1) What Turbo Prolog is doing with its memory management,
	(2) How to free the memory - ideally, a predicate like
db_garbagecollect for internal (rather than its current limitation to
external) databases would be needed - or just a single call that cleans
up all available heap and stack,
	(3) Why do I appear to be getting random numbers?

Could any gurus out there post to the email address below.  I'll
summarize and sent to the net the responses if there appears to be
sufficient interest.  Thanks in advance.

----------------------------------------------------------------------
Peter K. Campbell                      |
2/M6                                   |
Telecom Research Laboratories          | Phone : 03 541 6751
P.O. Box 249                           | Fax   : 03 543 6026
Clayton 3168                           | Email : p.campbell@trl.OZ.AU
Victoria, Australia                    |
----------------------------------------------------------------------