Subject: Re: A positive comment about 1.3 Sysinst
To: Dave Burgess <burgess@cynjut.neonramp.com>
From: Joel Reicher <joel@aaii.oz.au>
List: current-users
Date: 01/20/1998 12:28:50
Seeing as there's been all this desription about sysinst, I thought I'd 
mention something else. I've observed a problem on i386 at least with the 
disklabel calculations. I think it's due to the "skipping" of the first 
track on the i386 port, but could somebody who's done a standard install 
with sysinst please check their disklabel to see if their partitions are 
what they thought they were, and are on cylinder boundaries? I believe 
they're supposed to be on cylinder boundaries unless you decide to 
specify it all in sectors.

BTW, I've already send-PR'd this (4791).

	- Joel Reicher

On Mon, 19 Jan 1998, Dave Burgess wrote:

> > 
> > 
> > [redirected from port-i386 since sysinst is used on other ports]
> > 
> > The best thing is if you send a separate PR for each (and every)
> > failure condition that sysinst doesn't handle gracefully, like
> >   * trying to remount already-mounted filesystems
> >   * not unmounting sysinst-mounted filesystems when an upgrade/install is
> >     aborted
> >   * not noticing overfull filesystems,
> > 
> > etc, etc.  Sysinst detects and handles these as best we can, but it's
> > always good to have a list of more things to improve.
> > 
> > Thanks on behalf of Phil Nelson (who may not read port-i386) and the
> > rest of us.  BTW, I don't know about i386, but on other ports
> > sysinst's menu-driven disklabel editing and sanity checks should make
> > even more of a difference to full INSTALLs than it does with upgrades.
> > 
> 
> Three things:
> 
> 1)  Thanks for the help and support on getting something that works
> reasonably well.  It is an excellent start and something that makes the
> system far easier to install.
> 
> 2)  IMHO - We could really stand having sysinst worked as a separate product
> line instead of tying it to a version release for the OS.  Now that we
> have a working 1.3, wouldn't it make sense to get sysinst working with
> 1.3 and release a new "version of the week":, updating the program as we go?
> If that was the plan, the obviously this is a "Never Mind".
> 
> 3)  I have yet to actually have sysinst work with a SCSI drive and a
> 1542 SCSI card.  I've tried two different machine and ended up
> installing both by hand to get them so they could boot.  On both,
> sysinst points the boot to the wrong place on the drive for the
> system startup.
> 
> 
> -- 
> Dave Burgess                   Network Engineer - Nebraska On-Ramp, Inc.
> *bsd FAQ Maintainer / SysAdmin for the NetBSD system in my spare bedroom
> "Just because something is stupid doesn't mean there isn't someone that 
> doesn't want to do it...."
>