RE: Re: Re: Testing malloc results for null on the 400(was:HTTP Authentication / base64 encoder?)

> I know this is getting off of the subject, but I have not 
> heard this before. I don't understand why realloc is better
> than malloc in this case.

It's really down to allocation ownership - it's good practice in C/C++ to have 
the caller provide the memory, so that the onus is in the caller to clean up 
after themselves. In some/most cases the function has no way of knowing when to 
clean up ! This is especially true when the function resides, say, in a *SRVPGM.

> Also, I didn't verify this, but isn't realloc 
> significantly more expensive than malloc?

Slightly more so, as the pointer has to be checked (all though if the pointer 
is null, then it *should* be as fast as malloc() ) and if the allocation is to 
be extended, and won't fit in the current area of memory, then there will be 
the overhead of moving the current data at the old allocation to the new 
extended allocation.

--phil