methods@oasys.dt.navy.mil (William Mills) (11/16/90)
Turbo Prolog, Quintus, and perhaps other IMB PC implementations have memory management problems - presumably caused by the 640K ceiling - such as rapidly running out of stack space during recursion and a conflict between needing more code space and more heap space. Is there a solution using expanded memory? Alternatively, is there a Prolog that runs under OS/2 or a UNIX for the IBM PC that would avoid such problems by automatically accessing all memory?
todd@uhunix1.uhcc.Hawaii.Edu (Todd Ogasawara) (11/16/90)
In article <4361@oasys.dt.navy.mil> methods@oasys.dt.navy.mil (William Mills) writes: >Alternatively, is there a Prolog that runs under OS/2 or a UNIX for the IBM PC >that would avoid such problems by automatically accessing all memory? I see full page ads in AI Expert every month touting an OS/2 version of Arity/Prolog. I have no experience with this product though. Has anyone seen or heard anything about Arity's DOS-based Prolog interpreter/compiler lately? I haven't gotten any update notices or bug reports/fixes in two or three years....todd -- Todd Ogasawara, U. of Hawaii UUCP: {uunet,ucbvax,dcdwest}!ucsd!nosc!uhunix!todd BITNET: todd@uhunix INTERNET: todd@uhunix.UHCC.HAWAII.EDU
bradley@cs.utexas.edu (Fletcher Mattox) (11/17/90)
Current DOS products from both Arity and Quintus allow only 64k (1 segment) of stack space for each of the stacks--so, yes, they do run out almost instantly if you're passing around large terms. Arity's "solution" is their OS/2 product--which is utterly useless to those of us who know better than to move to OS/2. Quintus (actually LPA) has announced a 386 version of their DOS implementation, which will thus have no specific memory limits (it runs under a DOS extender, so can access all the memory in the computer). This, to my way of thinking, is the way to go. (Ken, is this product available yet?) Bradley ------------------------------------------------------------------------------- Bradley L. Richards "The abandonment of formal mathematics bradley@cs.utexas.edu is an extremely popular thing to do in uucp: cs.utexas.edu!bradley computer science." Dr. Robert Boyer, NACLP-90 -------------------------------------------------------------------------------
dowding@ai.sri.com (John Dowding) (11/17/90)
In article <14722@cs.utexas.edu> bradley@cs.utexas.edu (Fletcher Mattox) writes:
Quintus (actually LPA) has announced a 386 version of their DOS implementation,
which will thus have no specific memory limits (it runs under a DOS extender,
so can access all the memory in the computer). This, to my way of thinking, is
the way to go. (Ken, is this product available yet?)
Bradley
Is anyone going to come up with a good implementation that uses Windows
3.0 as the DOS extender, so I could run it on my grumpy old '286?
bradley@cs.utexas.edu (Bradley L. Richards) (11/19/90)
In article <DOWDING.90Nov16121021@palm.ai.sri.com> dowding@ai.sri.com (John Dowding) writes: > >Is anyone going to come up with a good implementation that uses Windows >3.0 as the DOS extender, so I could run it on my grumpy old '286? Note that Windows is *not* a DOS extender in the sense you mean. Windows provides a multi-tasking environment, as does DESQView (which in my mind is preferable to Windows). But either environment (or just standard DOS for that matter) will run programs which were written using a DOS extender like PharLap. I'm not exactly sure how many applications have been extended (in this sense) for 286's--the great advantage for 386's is that the extenders kick the CPU into the native 386 mode with a flat memory architecture. Presumably similar advantages can be had by using the 286 in true 286 mode, but I don't know if this is done very often since getting out of 286 mode literally requires a hard CPU reset (one of the reasons the 286 was called "brain-dead" by many reviewers). But again, DOS extenders are a completely separate issue from environments like Windows.... Bradley