Does TC's farrealloc have a bug?

[Paul Jarvis]

I have recently had a report of problems using huge far pointers and allocating
and freeing memory. The problem appeared to be that the free process did not
work hence memory got lost. The compiler was Turbo C (sorry version not known).
Changing to Borland C++ compiler with identical code - i.e. the same program
no problem. My conclusion is that there may be some problems with allocating
large memory chuncks but I sure don't want to try to prove it.
                           Paul
                        (pj at doc.ic.ac.uk)