> Given that this second stage tape booter is probably not _too_ bright (it 
> should be a variation on the boot program) it may be possible to kludge 
> this by using the 1.4.x booter, and the 1.5 miniroot to do the right thing.

I was thinking of trying that (I once did it on a SunOS sparc tape for
4.0 which had blown usr and beyond tarballs, using the 4.0 bits to install
4.1 tarballs).  That same approach worked in getting RT-4.4BSD up over an
RT-4.3BSD boot system I have.  Other than getting the right boot bits in the
right places, should it particularly matter which booter kernel is used?
If the 1.4.x works, keep it, until we get something better?

> Alternatively, we could probably hack the existing boot such that if the 
> boot device was 'Mxx' then instead of trying its sequence of netbsd images 
> (which won't work on a tape anyway) just load the 'next' image (which would 
> be the ram based miniroot) and start that up. That would presumably leave 
> the tape positioned past the install kernel and looking at the first tar set.

That sounds reasonable.  I like the idea of having a VAX tape set around
that does the complete tape install of all the non-X stuff.  It is slow
and kludgy, but it works.  If possible it should not require manual
installation of the sets from tape before continuing.

>  > And, how do I hook up a CD to an MVII?.... not so easy.
> 
> You user an RQD50 and an RRD40 :-)

The problem is WHERE to find such an RQD50 critter?  Does it work
correctly with an RRD42A?  Some of these bits are seemingly a
bit of unobtainium.....(:+}}... (especially at beer-bellie budget
prices....(:+{{...).

> >For the sake of discussion, and keeping the low-ram capability for
> >oldeVAXen around, should we rewrite a VAX specific low-end tape
> >install that is more traditional in nature rather than late-model
> >sysinstall based?  It would require we keep a second VAX-specific
> >tapeboot readme for just that instance.  But, I sense that a
> >minimal traditional install might be good.
> 
> This is essentially what I propose but you could still do some sysinstall 
> type stuff.

As long as it was minimal, and good to bring up at least the base/kern/etc
completely, that is what I was thinking.  Trim the sysinstall, but, since
we know it is only going to be for a tape install, trim the fat down to
the bare minimum.
 
> >Also, for the sake of discussion, can any sort of tapeboot run off
> >9 track reel scsi tape drives?
> 
> As far as the VAX is concerned if they are a TMSCP device they all "look" 
> the same.

I tried that one time and it did not work on my 3100 M/38 box.  Sadly,
I don't have a Qbus tape card for scsi 9 track Cipher drives,.... yet.
(Any cheap ones floating around in the bilges, do holler??????)
(I could also use the differential Qbus scsi tape card, but that
one is even scarcer, which might work on my IBM 9 track differential
deck.)

> >I have a couple of those that would
> >be nice to use, in lieu of a cd boot or a TK50 boot.  I am not sure
> >what differences in coding would be required, but it probably would
> >not be too much.
> 
> Might take a couple of tapes, the CD image is 53Mbytes

I have 6250bpi scsi drives.  That should handle 130 some odd mb of stuff
which would be plenty.  I will try your boot.img and see if that runs
up on them.  If so, I will cut a full tape for testing.

Thanks

Bob