🚀 go-pugleaf

In article <1992Oct7.001411.8605@novell.com> larrow@novell.com writes:
 > [...]
 > My question is why don't applications handle running out of memory in a 
 > nicer fashion? This may seem like a generalization, but it seems to happen
 > a lot. I would think that since the programmer allocates the memory via
 > a malloc or whatever, he/she would know when the memory allocation failed
 > and be able to recover gracefully. Is this a misconception? Am I thinking
 > on too small a scale?

If the application is using some form of malloc() to allocate memory,
then they are probably not checking the returned value for zero, which
(for better or for worse) indicates that the amount of memory requested
is not availble.  If it is not checked, then the assigned pointer will be
NULL.  Since DOS can't detect NULL pointers, then chances are it will be
assigned to, most likely modifying DOS's interrupt table.  Crash.

Most better-written applications have a shell around malloc() that
tests for a NULL return value and, if so, prints some informative
message and quits gracefully.

--
Elliot H. Mednick                                        P.O. Box 150
Wellspring Solutions                                     Sutton, MA. 01590
elliot@Wellspring.com                                    +1 508 865 7271