From owner-freebsd-hackers Sun Jul 16 20:34:30 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id UAA24409 for hackers-outgoing; Sun, 16 Jul 1995 20:34:30 -0700 Received: from haywire.DIALix.COM (peter@haywire.DIALix.COM [192.203.228.65]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id UAA24393 for ; Sun, 16 Jul 1995 20:34:20 -0700 Received: (from peter@localhost) by haywire.DIALix.COM (8.6.12/8.6.12/DIALix) id LAA07320; Mon, 17 Jul 1995 11:33:05 +0800 Date: Mon, 17 Jul 1995 11:33:05 +0800 (WST) From: Peter Wemm To: David Greenman cc: hackers@freebsd.org Subject: Re: utmp ut_host field In-Reply-To: <199507170207.TAA03570@corbin.Root.COM> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: hackers-owner@freebsd.org Precedence: bulk On Sun, 16 Jul 1995, David Greenman wrote: > >Hi all.. > > > >telnetd will store the dotted IP address in the hostname field if the > >actual name is too long to fit. > > > >w detects this and correctly looks up the real name. > > > >who does not, neither does finger. > > > >tcsh does really strange things.. it tries to > >gethostbyname("192.9.200.1") and other wierd things. > > > >I like the idea of using the IP address if the full name wont fit, but > >it's a bit annoying to see only some of the system understanding it. > > > >Would there be objections to me going through and fixing it? (ie: get > >rlogind to generate the IP address in the same way that telnetd does, and > >fix who,finger,tcsh etc so they understand it? > > Yes, I object. I would rather rip out the support in 'w'. This "feature" > was ill conceived and ends up making 'w' appear to hang all the time when a > nameserver is unreachable. This happens all the time on freefall and it is > more than a bit annoying. We've gotten bug reports about it too. I can imagine..... > I would be willing to compromise, however. If you would like to make the > lookup an option rather than the default, I would have no problem with this. > 'w' has a -n option to disable the feature, but I hate the feature being the > default. I have another alternative.. Would you be willing to allow specifing a really small resolver timeout for these commands? say 2 seconds? This would add only a couple of lines to the code, and would be a pretty reasonable alternative to the existing 75 second timeout. Something like this: #include if (!(_res.options & RES_INIT)) res_init(); _res.retrans = 2; _res.retry = 0; As for making it not the default, I'd be quite happy to do this myself if you'd let me use an environment variable to enable it for the utilities that care.. :-) (you know, like "BLOCKSIZE", which most of the disk utilities respect when reporting disk units (df, du, etc)). > -DG Cheers, -Peter