> Ok, so I've cleared up a bit of the mystery.
> 
> My particular interest was getting tapes to boot so that was where I 
> concentrated my investigation.
> 
> The directory src/distrib/vax/inst is where the file 
> 'bootroot-${release).fs' is supposed to be built.
> 
> The way it used to be built (in 1.4) was to take four files:
> 	edlabel (built in ./src/sys/arch/vax/boot)
> 	copy	(built in ./src/sys/arch/vax/boot)
> 	boot	(built in ./src/sys/arch/vax/boot/boot)
> 	xxboot	(built in ./src/sys/arch/vax/xxboot)

For the sake of discussion, you might want to look at the 1.4.1
tree.  That is the ONLY working 1.4.x level tapeboot I have found
(have tried 1.0, 1.1, 1.2, 1.3, 1.3.2, 1.4, 1.4.1, 1.4.2, 1.4.3, 1.5).
That one was done right, and uses the newer sysinstall scripts.
The 1.4 and earlier used the old method (which is good, but edlabel
is a real buggerbear).  For speed, the vanilla 4.3BSD code can't
be beat.  Strange as it seems, that loads 10x faster than NetBSD
code.  It might be worth emulating, because of its speed on TK50
tapes.  The 1.4.1 code works fine but is painfully slow on TK50.
Whatever 1.4.1 is using, we need to use in 1.4.4 and 1.5.1, maybe?
The 1.4.3 bootfs almost works, but can't mount anything in my hands,
for some reason, so the install croaks (permissions?).

I'm a genuine dummy in this area, so other than having tested all
the different bootfs's to see what worked, I don't have any answers.
Ragge, maybe?

I like the old way, myself, for simplicity, but use the 1.4.1 bootfs
mostly, because it never fails me.  Am I the only joker still doing
tape boots?

Bob