On Tue, Mar 04, 2003 at 12:37:27AM +1030, wulf@ping.net.au wrote:

> -- Unlike the Boyer/Moore and the other cited algorithms, memmem(3) has
>    little overhead and may actually yield better performance under
>    certain conditions.

Uhmm, I can't let this slip through uncommented.

Sure, for single calls per pattern in small "haystacks" the brute force
implementation is the best. In general it certainly is not.

I won't comment on unecessary bloat in libc, since we already have bm(3), and
this is very lightweight compared to it.

But please at least put a prominent pointer to bm(3) in the man page.

Martin