Hello, probably a stupid remark, but I'll do it anyway :-)
Let's assume there are some contexts where doing a memory allocation with a size of 0 is invalid while in some other contexts it is valid. Wouldn't it make sense in that case to have two functions that do memory allocation? Of course such a split would require full review of existing code and might introduce complexities for instance for realloc (was the original memory block allocated with a possible size of 0 or not?). Laurent