🚀 go-pugleaf

In article <C05Bno.9K@boi.hp.com> verghese@boi.hp.com (Phil Verghese) writes: 
   Here's some info from the brochure:
       PocketSoft  1-800-826-8086

   VMData is a replacement for malloc.  When you want to dynamically
   allocate data, you call VMData functions instead of malloc.  The
   virtual storage can be in EMS, XMS or disk.

   There's a version for DOS and Windows.  Each version has two
   libraries: one for development (debugging) and one for production.
   The development library does more error checking than the production
   one.

   Languages/Compilers supported: MSC 5.1, 6.0 and C++ 7.0; Borland C++
   2.0, 3.0, 3.1; MS Fortran 5.0,5.1

It is easy to write a program which would try to malloc space in memory and if
that fails, open up space on a disk.  The problem is getting at the memoroy on
disk. Consider a linked list:

 elem_1 -> elem_2 -> elem_3 ....

where elem_i is a struct of type elem_type. 

Suppose elem_1 is in memory but elem_2 is on disk.  Let p be:

elem_type *p;

Then if p = elem_1, *p is easy since it is in memory.   Then what about the
statements:

p = p->next;  ...*p;   ???

Somehow the virtual memory package must interact somehow with the compiler or
must have some sort of wrapper routine in its library which forbids
statements like:

p = p->next;

forcing one to use a library routines such as:  p = vm_next(p);
and vm_star(p);

In any case I will call call the company on Monday because I am quite
skeptical.






--
     _____________
#___/John E. Davis\_________________________________________________________
#
# internet: davis@amy.tch.harvard.edu
#   bitnet: davis@ohstpy
#   office: 617-735-6746
#