On Sat, 18 Dec 1999, Bob Nestor wrote:

> Frederick Bruckman  (fb@enteract.com) wrote:
> 
> >On Fri, 17 Dec 1999, Bob Nestor wrote:
> > 
> >> The one remaining problem with sysinst seems to be an MI one.  There are 
> >> times that it gets confused about the updated disk label and confuses the 
> >> memory-disk with the physical disk. When this happens the newfs step 
> >> fails and thus the installation fails.
> >
> >Demanding a reboot is crude, but it would address that, too.
> >
> True, but I think if you're going to do something like this you need to 
> actually force the system into a reboot rather than ask the user to do 
> it.

Yes, of course. I agree completely.

> I really think rather than doing this you should consider either _NOT_ 
> building and distributing an Installation Kernel with 1.4.2, or 
> installing the one line kernel patch into your Installation Kernel.

"make release", in current, now builds and installs an install kernel.
If it's broken, it'll have to be fixed. I haven't yet requested a
pull-up mainly because I don't want to deal, at the last minute, with
breakage due to the obj dir/OBJMACHINE stuff. [I developed parts of
the make goo on an i386 with MACHINE set to mac68k. The final test was
done on the mac, on the release branch, but it doesn't cleanly update
to current; the patches actually committed were hand made. LOL!]

I feel the install kernel is useful in it's present state, warts and
all. As long as you have a third party partitioning utility (as
present mac68k users must), and as long as you can arrange to make the
sets accessible (ftp or nfs or a scratch partition), you can install
faster than with the Installer, and you can dispense with Mkfs
altogether. I've warned people about the partitioning, to keep down
the redundant PR's, but all the other bugs may as well be known.

I'll respond to some of your other points in private mail.