Well, for "historical" mac users, the main difference is between MacOS
9 style processes versus OS X.
On OS9, your process had a spot in memory, and it could change from run
to run (specially with vm enabled, where some of the "free" space was
allocated dynamically to other cooperative processes, but let's not go
too far in the last millenium ).
On OSX, your process lives in its 4 gigs of virtual space on our aging
32 bit systems, and pretty much everything you see that takes place in
the process itself is very deterministic in terms of memory
allocations. Your stack starts at the same address by convention
determined by Apple, as well as the standard heaps used by the malloc
subsystem. So no, it is not strange to see behaviors to be near
identical across runs, across machines, and in many cases across OS
versions. All due to the nice shielding of the virtual address space.
However, any process with a UI typically gets a good randomization for
memory allocation, even with only the raw allocations done by the
underlying frameworks (Cocoa, Carbon, you name it) that happen to live
in user space.
Ludo
On Vendredi, octo 22, 2004, at 22:04 Canada/Eastern, David Duncan wrote:
On Oct 22, 2004, at 09:45 PM, Gohara, David wrote:
Thanks for the responses. It would appear that it is indeed a
case of malloc. One thing I noticed and I've never done the test
before is that the same code compiled on a different machine gives me
the exact same memory addresses. For example, if I compile and run
the code on my G4 laptop or G5 desktop. How can that be? I may be
misunderstanding something with regards to memory addresses but
shouldn't it be very unlikely for memory allocations to be identical
on two different machines?
As long as the machine code that is run is the same (same gcc version
and compiler flags, same source, etc) then there is no reason for
anything to change. After all, malloc() an algorithm to allocate
memory -- it would be very strange indeed if it produced different
results on the same (machine) code with the same starting point.
--
Reality is what, when you stop believing in it, doesn't go away.
Failure is not an option. It is a privilege reserved for those who try.
