Subject: Re: RFC: route(8) host/bits vs. net/bits
To: Brian Ginsbach <ginsbach@NetBSD.org>
From: Christopher W. Richardson <email@example.com>
Date: 05/12/2005 14:10:42
-----BEGIN PGP SIGNED MESSAGE-----
"Brian Ginsbach" <ginsbach@NetBSD.org> writes:
> On Thu, May 12, 2005 at 11:09:32AM -0400, Christopher W. Richardson wrote:
> Would you agree that route should at least honor -net 172.31.73/24?
> (See attached patch.)
>> Probably the correct change is for route to display
>> 172.31.73.0/24, so that it is clear. If I type in 10.1 as an
>> address, I expect that to be interpreted as 10.0.0.1.
> No, I believe netstat(8) would need to be changed.
My mistake; I meant netstat, not route.
> As you suggest 2.0 and later route(8) already DTRT when it
> displays such a network address. I've been looking at a 1.6.2
> system so maybe netstat has changed as well but from looking at
> the current code I'd say no. Compare netstat/route.c:netname()
> with route/route.c:netname() and note that the latter no longer
> shortens networks to the "correct" number of quads.
Yeah, I don't think either one has changed ... which gets back to
POLA, and says maybe we shouldn't change anything since, as far
as I know, netstat(8) and route(8) have behaved this way ~
However, it does make sense to me that route(8) should honor the
combination of -net and the "/" semantic the same as if the
- -prefixlen argument had been given, and at quick first glance it
appears that your patch does that, while maintaining the -host
"/" behavior, so I'm OK.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (NetBSD)
Comment: Processed by Mailcrypt 3.5.8 <http://mailcrypt.sourceforge.net/>
-----END PGP SIGNATURE-----