Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!sdd.hp.com!hp-pcd!hpcvra.cv.hp.com!billw
From: billw@hpcvra.cv.hp.com. (William C Wickes)
Newsgroups: comp.sys.handhelds
Subject: Re: ROM versions and upgrades
Message-ID: <25590099@hpcvra.cv.hp.com.>
Date: 4 Feb 91 22:48:35 GMT
References: <633148B940000929@gacvx2.gac.edu>
Organization: Hewlett-Packard Co., Corvallis, OR, USA
Lines: 57


Mark A. Ordal writes:
> While on the question of ROM upgrades, I want to mention what I
> heard last March when I bought my revision A machine.  My dealer
> heard that 500 people were given free, pre-release, HP48's around
> Thanksgiving of '89.  These folks were (apparently) the people posting
> extensive features lists the same day the 48 was officially released.
> 
>   Does anyone know if this is true?  If true, were those machines
> revision A?
> 

Actually, five (5) people were given rev. A HP48's on about February 1, 1990,
one month before intro.  These were in fact the people that reviewed the
calculator at the release date.

> So my questions really are:
> 
>  1) Why were revision A machines sold at all?  The usual purpose
>     of this kind of pre-release testing is to kill as many bugs
>     as possible before the software (or firmware) is actually
>     released for sale.

As any credible software developer will tell you, no software of the
complexity of the HP48 can ever be certified as bug free, no matter
how much testing you do.  Products are released when testing data
predicts that it is unlikely that remaining bugs will be so numerous
or so severe that the products will be unacceptable to customers.
Unfortunately, we missed the matrix inversion bug, through a
combination of human error and bad luck, and it was not discovered
until late January, after many thousands of ROM's (A, B and C) were
built.  There was little point in scrapping those parts and delaying
the introduction by several months, given the possibility of more
defects to be found.

>  2) Why weren't competent testers chosen?  Or if they were
>     competent, how is it that so many ROM versions were made
>     without killing bugs?  It took until version D to kill the
>     matrix inverse bug (which fortunately had a simple work
>     around) and until version E to kill the complex number bug
>     (which has NO WORK AROUND!).
> 

The complex number bug was not found because it was so obvious--as Sherlock
Holmes might say, "you saw, but you did not perceive."  And there certainly
is a work-around--enter (theta,r) instead of (r,theta).  Hardly a showstopper.

> I have to wonder if management wasn't applying extreme pressure
> to get the product out the door and onto dealer's shelves.

Show me a product for which this is not true.  I certainly wish my PC software
were as defect-free as a rev. A HP48.  People who can work endlessly
on something until it is absolutely perfected generally turn out to be
financed by people who don't have that luxury.

Bill Wickes
HP Corvallis