Subject: Re: CVS commit: src/distrib/utils/sysinst
To: None <david@l8s.co.uk>
From: None <cgd@broadcom.com>
List: source-changes
Date: 07/21/2003 15:24:27
At Mon, 21 Jul 2003 21:52:45 +0000 (UTC), "David Laight" wrote:
> IMHO the latter is clearer, the test is 'is this string null' not
> 'does it have zero length'.

Can you point me at a definition of what a "null string" is?

empty string?  sure.  you test that.

NULL string pointer.  not what you're checking.

String starts with a NUL character?  you test that, too, but that's
the same as "empty string" if you're talking about C strings...


> In any case, even on i386 gcc -O2 generates:

is that a modern gcc, or the ancient one NetBSD ships?

Don't use the latter to justify making tiny "optimizations" to code at
the cost of readability or meaning.  Two wrongs don't make a right.



cgd