Evaluation of expressions involving string literals

[Cameron Paine]

I'm porting some code to an ANSI-conforming compiler and have encountered
an anomaly which my reading of K&R(2) and H&S(2) hasn't resolved. In the
original code, I used many expressions (the reason for doing this is not
relevant) such as:

	char *cptr;
	cptr = "XYhello, world!" + 2;

which has traditionally rendered:

	*cptr == 'h'		/* because cptr => "hello, world!" */

Under the ANSI-conforming compiler, I find that:

	*cptr == 'X'		/* because cptr => "XYhello, world!" */

Experimentation has revealed that:

	cptr = "XYhello, world!";
	cptr += 2;

achieves the result I'm after but this is of little use to me.

The question then is whether the new compiler's behaviour is correct. The
documentation I have implies that it isn't but I'd like to hear what the
lawyers think.

Thanks.

Cameron
-- 
cbp at foster.avid.oz - ACSnet
cbp at foster.avid.oz.au - Internet
...!{hplabs,mcvax,nttlab,ukc,uunet}!munnari!foster.avid.oz.au!cbp - UUCP