Evil memory problem

Assuming you''re using C++ have a look at the memory manager here (near the bottom it''s called MMGR)

You can use it to track all allocations and deallocations in your program and it can give you an indication of when memory is corrupted. If you''ve got it tracking everything then it can make memory allocations and deallocations rather slow but the information it provides can be highly useful.

softice anyone?


If God played dice, He''d win.
—Ian Stewart, Does God Play Dice? The Mathematics of Chaos

Maybe you could try running the project in a friend so you make sure you don''t have bad memory chips on your computer?
In my experience (it''s not much, but hey...):
maybe it''s something that you''ve overlooked and no matter how long you stare at the program you won''t see it. Try taking a break for a couple of days.

when you pin point the solution (or at least the part you feel is suspicious) try "taking it outside". This is my term for creating a smaller simpler project that just tackles the problem you have, out of the chaos of a large project with lots of source files.

Last word, since the variable is going corrupt, are you sure you don''t have an array that goes astray anywhere in there? Try looking at the base adressess of your variables and think if the adjacent variables have correct adressess according to the type size. For example, if you have a float and a short int that have a memory address of 0x000f348a and 0x000f348c respectively, then probably something is wrong (since a float takes 4 bytes of space).
Good luck man!

First of all, are you compiling with all warnings turned on? Sounds like you're writing past the end of an array maybe? This seems to be the perfect time and place for a shameless plug:

http://www.gamedev.net/reference/programming/features/modcpp/

[edited by - dmikesell on June 3, 2004 12:11:01 PM]

Thanks for the assistance so far Still haven''t located the bug though

Monder: I am using the memory manager you pointed out, sadly it sometimes works - sometimes doesn''t. When it does it seems to be fairly misleading in this case.

daveangel: Softice looks expensive

moucard: I''ll prolly take a break, its been getting pretty intense lately. I am thinking it could *possibly* be a memory chip issue (Windows has been getting less stable recently). I''ll put Visual Studio on another machine, and check it out there.

dmikesell: Thats what I was thinking. I have all warnings turned on, but my code is fairly robust because I usually heed those early on. I remember reading your article a while back and promptly changed styles to suite. STL is now featured very heavily - in fact it was changing to STLport that originally demonstrated this error. Now I''m switching between the two STL implementations (VS.NET & STLport) - both giving different errors