MEMORANDUM

TO:     Joe Dreussi
FROM:   Keith Corbett
DATE:   2/24/87
SUBJ:   Account Issues In Sarah Long's Area (ref: her memo 2/16)
COPY:   A. Salamon, C. Bee-Latty, D. Lee, B. Deffenbaugh,
	B. Campbell, C. Regis, T. Hatch, R. Genest

This memo is to follow up on Sarah's outstanding customer issues.  I
concur with Sarah's assessment of these customer situations, but in some
cases I think Sales and Service, when we arrive at solutions, might just
be looking at things from different perspectives.

1. [ADA port for Capt. Wood] We will do our best to respond to any
Window System questions coming over the hot-line.  This is one of our
real weak areas; since Debra left, I have often relied on you and the
tech reps for help with complex window system problems.  But, I'll be
happy to lend any support I can.

2,3,4. [Iris 2400s for WPAB] No S/W support required here.

5. [MicroCompiler for Capt. Vermillion] Sarah's solution -- letting him
use the beta version, cautiously -- should actually render his support
contract null and void.  There is no way we can support someone over the
telephone when they are using a broken, dangerous utility like the MC.
Note that there is no one left at LMI, with the possible exception of RG
or GJC, with the knowledge to support MC.

From our point of view, it's not just a question of being able to answer
his questions on the MC -- we will never know what weird state his
software environment is in!

This is still an Engineering issue -- they committed last summer to
assign someone to make the MC really work and document it properly.
Failing that, we shouldn't support it in any way.  I expressed this (not
so vigorously, perhaps) to Sarah when it came up.

We will try to keep up with Capt. V. regardless, but this is going to
bite us someday.

5,6,9,10,11. [Vista/Iris for various people] I will be circulating a
status memo on this release shortly; the bottom line is that I have
received (2/20) Engineering's final tapes, and it needs a final "sanity
check".  The tapes I got today have pretty drastic changes from the
"beta" tapes.

I think we can re-install and minimally check out both Vista and Iris in
about 2 weeks.  Assistance with this -- installing it on your system, or
a demo system, for example -- would be greatly appreciated.

After that, aside from duplicating tapes for Inventory, our part will be
done and the materials will be in Manufacturing's hands, and should be
pursued with them.

7. [CDC 300mb system from Rockwell Rocketdyne] Ken Dehoff called this
week to tell me that the System 5 installation on this system was a
complete disaster.  I don't understand how; Wendy Rowe assured me, in
writing, that she had checked this installation configuration, and "had
no problems".

The real surprise came when he told me they only have *one* 300mb drive,
for a 2x2-Plus!  We all know that configuration really doesn't work...
there isn't enough space to fit a comfortable configuration.  System 5
made the situation much worse (LOAD ate up the PAG1 partition, and he
can only run a single Lambda).  Ken wouldn't give me the current layout,
or I would tried to see if there is any better layout.  He is too p***ed
off at LMI to deal with these issues anymore; quote: "I have other work
to do".

I firmly believe, if they are truly running a 2x2-Plus with one 300mb
drive (and Sales should confirm this), that we should attempt to sell
them a second drive.  Heck, we're already having a garage sale on
everything else.

[Perhaps it's time to sit down again, Sales and Service, and talk about
minimum sellable vs.  shippable vs. workable configurations.]

8. [3.0 for Kodak] Many customers would like to know where their 3.0 is.
It's not hiding somewhere.

Inventory is down to a handful of tapes, and the VAX was fixed just a
few days ago (and it's still not stable).  It will be 2-3 weeks before
kits go out in any volume.  [My best projection on volume would be 5-10
kits per week, but Tom Hatch or Dan Remeika could give us a better
estimate.]

We still need 300+ new tapes -- using scrap tapes has caused a lot of
aggravation, for me, for the field, and for the customers, but that's
all I have to work with.  LMI apparently doesn't have the $$$ to send
the tapes out for duplication, which would be the sensible thing to do,
given the delays caused by lack of tapes and VAX problems.

Meanwhile, I will add Kodak to the long, growing list of "top priority"
3.0 shipments.  That list should be reviewed by Sales to re-prioritize,
or it will grow to the length of the customer list.  I hope to circulate
another memo soon with the shipment lists and backlog, but I need a
complete list from Manufacturing on what they have shipped since 12/1.

12. [A/D driver for NASA] No S/W support required, or possible.

I would like to point out that our experience with Lambdas running
custom Multibus boards (limited so far to Flavors' and Jim Avery's, the
latter resulting in a system return) has been fairly negative.

Admittedly, RTime is better equipped to deal with this, but the same
problems do apply here. The Lambda is too fragile, especially Unix.
This will require a custom version of System 5. Contractually, it
requires a special waiver of responsibility for system support and
maintenance, which spells out the customer's obligation to reinstall
vanilla software before we can handle problems in H/W or S/W support.
June Morris drafted the document (an attachment to the standard system
service agreement) that Ron Rando asked MCC to sign; by the way, they
never did agree to the restrictions we require.

The problems with delivering this driver could almost certainly occupy
almost $30,000 worth of LMI employee's time and aggravation.  Jim
Avery's multibus problems took up much of Don McLain's time, and a lot
of mine, for several months.  Note that Don's time was relatively cheap
(in comparison to supporting other sites such as NASA), since we didn't
have to fly someone out there every day the system crashed.

George Carrette might have eventually gotten the Votan interface to
"work", but the Lambda couldn't even read from this board without
crashing, and nobody in Engineering was able to find a way to reserve
more than 8K of address space.  Naturally, the device driver
documentation, such as it was, is totally obsolete for System 5 /
Release 3.

Yes, this "deal" should be followed up -- by a cost/benefit (and maybe a
time/motion) study on all the special deals LMI has agreed to as
incentives for buying Lambdas.

13. [chaos routing for Thinking Machines]  Thankfully, this is noted
before the final patch release. I will let Robert Putnam know about
this; if it's safe to expand the routing table, this should be done as a
standard patch.  That makes it possible for a new release to come out
without blowing the customer away.


KMC