Hello Darin!

02 Sep 03 21:31, you wrote to me:

 DM>>> Yes.  But not in C.  Okay, that's not entirely true - you'd have
 DM>>> to make significant changes, which are not always an option.  If
 DM>>> you wrote both the caller and callee, then these changes are
 DM>>> what I'd suggest anyway, even in C++.

 BS>> Hmm.. I looks like it would be easier with the extra pointer.

 DM> Personal preference, I suppose.  However, the method I proposed below,
 DM> despite being "more difficult", is, I believe, usually the better
 DM> choice.

 DM> 1.  It is fast.  With no malloc/free (slow, expensive function calls),
 DM> it performs as fast as the caller allows.  The only thing that would
 DM> be faster is if we didn't need to copy at all, e.g.: char const*
 DM> test() { return "test"; } .  There are some other
speed-ups possible
 DM> (e.g., caching the length of the string being copied so that the next
 DM> time we're called, we don't need to recalculate it), but overall it's
 DM> very fast.

Yes that's right! A const char*, should not require memory allocation afaik?

 DM> 2.  It is complete safe.  Since we check the size of the incoming
 DM> buffer, we know we won't overwrite anything.

 DM> And, of course, it's a bit more than needed.

Yes.

 KK>>>>> #include 
 KK>>>>> #include 

 DM>>> char* test(char** buf, size_t buflen)

 DM> We don't need the extra * on buf throughout here.  So this would be
 DM> char* buf ...

I saw so, but I forgot to ask.

 KK>>>>> {
 DM>>>   /* char* tmp = (char*)malloc(80); */
 DM>>>   /* if(tmp) sprintf(tmp, "test"); */

 DM>>> char* source = "test";
 DM>>> if (buflen > strlen(source))
 DM>>>   sprintf(*buf, source);

 DM> And this would just be buf, not *buf...

Indeed.

 DM>>>   /* return tmp; */
 DM>>>   return *buf;

 DM> As would this ....

Well it doesn't matter? It would return the address anyway.

 KK>>>>> }
 KK>>>>> int main()
 KK>>>>> {

 DM>>> char buf[16];
 DM>>> printf("%s\n", test(buf, sizeof(buf)));

 KK>>>>>    fgets(buf, 16, stdin);
 KK>>>>>    return 0;
 KK>>>>> }

 DM>>> No malloc/free's here.

 BS>> No i see thank you :)

 DM> No problem.

:-)

Regards,
Bo


--- Squish/Linux v1.12.002 BETA