[424] in NetBSD-Development
Re: NETBSD
daemon@ATHENA.MIT.EDU (Greg Hudson)
Wed Jan 18 10:22:27 1995
To: yoav@MIT.EDU
Cc: netbsd-dev@MIT.EDU
In-Reply-To: Your message of "Wed, 18 Jan 1995 04:13:36 EST."
<199501180913.EAA00733@I-See-Everything-Twice.MIT.EDU>
Date: Wed, 18 Jan 1995 10:22:08 EST
From: Greg Hudson <ghudson@MIT.EDU>
> A) ttyv{1-6} aren't created, I had to MAKEDEV them by hand
> (and MAKEDEV all didn't create them).
sdfranks reported the same problem; however, every time I install the
system, my script creates hese devices. I suspect that there are
multiple versions of the installation disk, and one of them creates
/dev/lkm (which my script uses as an indicator that it's already run
that step).
I will modify my script to always create the devices, since it's an
idempotent operation.
> b) kerberos is REALLY slow. There is about a fifteen second delay
> between me typing kinit yoav and the program giving me
> a password prompt (it does get loaded, since it gives
> me the Cygnus message).
I have the same problem on my machine, but not on granola. I don't
know what the problem is. My guess is that something is causing name
resolution to be very slow.
> c) login is broken. It seems somebody blew away my login with
> another, and it fails to work.
Nope, my script leaves /usr/bin/login alone for now. Perhaps you were
expecting it to be Athena-ized?
> e) some stuff is still wrong with shells, it could be the system
> wide cshrc or /bin/athena/tcsh or something.. I'm
> still working on that...
Some notes:
* The system-wide cshrc was reinstalled when I installed xdm.
This seems to have reverted a number of people's changes,
because they thought that modifying the installed files was
a meaningful way of fixing problems. Let me emphasize once
more: make your changes in the source tree, or don't expect
them to be persistent.
An awful lot of things broke when I reinstalled a number of
Athena packages because of people modifying installed files.
This included the path in the system-wide cshrc file and the
shell run by the default xsession script. I would say that
this breakage reflects a bigger problem with the management
of this port than the problems jhawk pointed out with
building various packages in /afs/sipb/project/sipb-athena.
* pmail is in the sipb locker, not in /afs/sipb/project/sipb-
athena, so jhawk was a bit confused about his flame.
Also, I think Yoav was referring to the original person who
built pmail, not someone who rebuilt it. I have modified
the source tree for popmail to have the right paths for
NetBSD, so that things should hopefully work in the future.
* I do not have a great idea off the top of my head for a way
of updating files on people's local disks which might break
when we change things in /usr/athena. Also, I can't do it
unilaterally, since we would have to agree on a versioning
scheme by which we update, say, /usr/athena/etc/version
whenever we change something that's likely to break the
local filesystem.
* Any update procedure should probably not involve my
installation script. An update script and an installation
script just don't share very much in common. Also, updating
can leverage off of AFS, while installing can't. For
instance, the update program (or, more generally, the
private workstation management program) can be called
/usr/athena/bin/asetup and people won't have to ftp it in
order to run it.
