JB> char array[20]="hi";

 JB> will compile differently to

 JB> char array[20]; strcpy(array,"Hi!");

 RS>    Yeah, it will, because the first allocates 20 bytes, and sets the 
 RS>    first three bytes to 'h', 'i', and 0, respectively, whereas the 
 RS>    second just allocates 20 bytes, and waits until strcpy() changes 
 RS>    them.

More than that, the first one will allocate 20 bytes, set the first to 'h',
the second to 'i', and the rest to nul.  The second one will leave the
entire block of 20 characters alone, and then proceed to change the first
three, without changing the last seventeen.

 RS>    My point wasn't this; it was rather that the NUL is tagged onto the 
 RS>    ends of strings if you initialized them when you declared them.  
 RS>    For example, 

This is not entirely true.

char array[2] = "hi";

This you think should take three bytes - but ANSI C allows this particular
construct to take the two you asked for.  You're asking for trouble if you
try to use printf, strcpy, or any other "string"-like function,
on it.  However, accessing array[0] and array[1] is perfectly fine.

 RS>    even though there isn't room.  You're right, a good compiler may 
 RS>    issue a warning here, but that's about it.  If you run this code, 

Nope - a good compiler should say nothing.



---