>	I have done everything that I can think of concerning and have 
>covered every possible variable concerning partitioning and installation. 
>Then why:
>
>1) the active partition data shows up in the LAST entry instead of the first?

1) The installation uses the last partition because of the
possiblity that you may have other OSs install in the first three. 
Also, traditionally the boot blocks reside in partion 'd' of the
disklabel, i.e. partion 3 = d if logically numbered.  This really
isn't a problem, in fact I like it.

2) The default 24MB partion gets installed by disklabel because the
offset of the BSD partion is set to 0 and the 1st stage boot loader
contains the default partion table (MBR) gets installed here.
This can be circumvented by setting up the 'a' and 'c' partion
offsets to 1 and 'd' set to use the entire drive.  Never install
the 1st stage loader via disklabel.  Well, if you don't have
bootstrap code in the MBR, from a multi boot loader or something
else, go ahead and install it but be prepared to rerun fdisk to
restore the partioning scheme.  If you could skip the equivalent of
DOS's idea of the first track you would be even better off, although
I couldn't get this to work.

Number 2 is really a big problem.  OSs on PC's that rudely use the
first track can damage BIOS extension code (ProDrive?) that may
reside there and BSD may be damaged by other OSs as some expect the
first track to be empty. We need to get this fixed ASAP.

It should be noted that when I say the first track, I mean the first
track as defined by DOS geometry.

You can live with these two problems though.  Just don't use any BIOS
extenders and never let another OS munge with the partition table.

>2) I am given some 1023 cyl 255 hd 63 sec nonsense instead of 902/8/46 as 
>the CMOS and boot interrogation agree.

You are seeing BIOS/DOS/Net{Open}BSD differences in the way drive
geometry is calculated.  Partion tables should always be built
with the same geometry that DOS uses.  If it isn't or is mixed
it can cause catastrophic problems with other OSs that may reside on
the machine.  I had NT trash my BSD filesystems because it thought
there was free space on my BSD drive when in fact there wasn't.  All
major OS's use DOS geometry in the partion table and native OS
geometry for their normal operations.  To be PC compliant and
friendly, we should do the same.  Could we please get this fixed.

Disklabel needs work too as it should fixup the partion table to the
state it was in before it installed the first stage loader.

You can live with this as well.  Just use the values that are
displayed at boot time.  Again, no BIOS extenders and stay away from
the 1st stage loader unless needed.

>3) every time I let the kcoth11 floppy proceed without the installer 
>disk in hopes of reaching copy_kernel, it gives me fd0 timeouts and a reboot?

I don't know. Could be a bad floppy, flakey hardware, some
incompatibilty between your machine and the driver, or just
something wacked with kcoth11's install procedure.  hmm... Need more
data.

>4) Would the NetBSD community PANIC by allowing someone to do the
>straightforward and VERY politically incorrect: 

># mount -w /dev/wd0a /mnt2
># cp /netbsd /mnt2  (from a proposed third floppy or fifth .fs file 
>called 'kerndisk-a' or 'kerdisk-o')
 
>I would consider this a BUG!

I don't see your point on this one.  Please explain.

We really need to fix these types of problems and ensure that the
installation docs cover these types of topics.  OS/2 2.1 had about
30 pages dedicated to partioning schemes.  Lets do something similar.

good luck,
erik