compare strings, strcmp

[Dan Mick]

In article <5205 at mentor.cc.purdue.edu> aic at mentor.cc.purdue.edu (George A. Basar) writes:
|
|In article <308 at charyb.COM>, dan at charyb.COM (Dan Mick) writes:
|>In article <11605 at smoke.BRL.MIL> gwyn at brl.arpa (Doug Gwyn) writes:
|>>#define  StrEq( a, b )  (*(a) == *(b) && strcmp( a, b ) == 0)	/* UNSAFE */
|> 
|> Why the UNSAFE comment?  This looks like utterly standard C to me...
|> -- 
|
|  It is not unsafe, it is just that he was looking for a a way to not
|perform the strcmp if the first chars were unequal.  He stated it was
|unsafe(performance-wise) since the order of evaluation is suspect. So the
|strcmp may be performed before the test for *(a)==*(b).
|  A more 'safe' StrEq is
|
|#define  StrEq(a,b)  ((*(a) == *(b))?strcmp(a,b):*(a)-*(b))
|
|  This will guarantee order of evaluation, but with the added overhead of
|the terniary operator. Actually, the difference is only 3 instructions in 
|favor of my version(22 to 19).
|

What?

First of all, I'd think Doug knows whether or not he meant "unsafe"; after
all, the comment is pretty unambiguous.  Secondly, how can there be any
other evaluation order, considering the precedence and the short-circuit
behavior of &&?

This looks suspiciously like "Here, here, I've got an answer" but it's
*wrong* and *unhelpful*.

Sigh.
-- 
.sig files are idiotic and wasteful.