Re: CVS commit: src/lib/libc/inet

[Brian Ginsbach <ginsbach%NetBSD.org@localhost>]

On Sat, Jan 19, 2008 at 10:07:55PM -0500, Perry E. Metzger wrote:
> 
> Brian Ginsbach <ginsbach%NetBSD.org@localhost> writes:
> > Never said it did.  If the goal is to keep it looking as close to
> > the bind source then we really should remove itojun's change from
> > the libc version.  It is now redundant code.
> 
> I doubt it causes any sort of performance harm, and the
> belt-and-suspenders aspect of it is comforting...

In use, probably not, but it is an unnecessary redundant check.
It will have some performance impact.  I spent several days chasing
one of these sort of things at the real job.  Code was changed on
a platform where it _probably_ didn't have much of a performance
impact.  It was then used on a platform where it had a huge impact.
It was difficult to track down because the new code looked "right
and comfortable"...

Plus I thought the NetBSD way was to do the "right thing". :-)