Stigers, Greg ~ AND wrote:

> But having recently skimmed thru Being Digital again, I wonder if this
> measure, years of age, is relevant. There are so many other factors to
> consider, much the same problem we have with benchmarks. How do you
> measure an OS? Years of existence just doesn't make sense.

Look at IBM/CMS under VM/ESA.  It can still run a "virtual" model 360
under DOS [and we ain't talking about the PC variety] (with that
possibly emulating a 1401, etc).  Not just age.

But then again, we have Pick, some esoteric Tandem, Cray, Cyber, Univac,
etc., operating systems with a short lifespan.

> And one wants to avoid an IBM k-loc (thousands of lines of code)
> mentality. Hours of useful production time makes sense for maturity,
> but is that per server, or per user?

If so, today's standards stink by nostalgic measurements.  The first
computer that I was intimate enough with to generate a new nucleus
(a.k.a. modern 'kernel') was IBM 360 DOS, and the nucleus was under 4K.
Yes sir, the resident OS was 4K.  But this waxes nostalgic, I'll stop.

Subsequent systems, additional requirements, but how many reports have
been carried over from the old batch card reader/line printer days?
Or perhaps enhanced with HTML tags?

The killer plague seems to be the GUI craze, which (sorry Alfredo) made
me dislike the Apple Lisa and the follow-up Mac.  At the time they were
pressing the limits of price/performance and memory/disc requirements
to deliver basic functionality, but somehow they managed and the GUI
became commonplace.  I'm not "totally opposed" - it brought computing to
the non-computer-guru.  It just seems that much of the complexity, cost,
and focus of current computing lies in it's "appearance" and not it's
function.  Gigabytes of memory and disc and millions of lines of code
to develop the ultimate "Hello, world" program with an animated spinning
java globe and a waving hand.

Not trying to start any cross-platform or philosophical debates, just
venting a bit of steam that having someone's picture ID in the database
starts to outweigh calculating the proper pay correctly, on time,
everytime.  Maybe it's time for another "paradigm shift" back to [data
processing] family values.  (I swear not to use those phrases again this
year :-) ).

> Do mips, flops, or tps matter? Something like mean time between
> failures does matter to many of us.

TPS certainly count unless you're in rocket science.  MTBF should count
irregardless.

> OTOH, what has your OS done for you lately? As for me, Alfredo's
> success stories are pretty persuasive. They answer the question, "but
> what can it do?". MPE runs. MPE has come a long way, and offers
> something rare:
> it's own robustness combined with adherence to a quite different
> standard, POSIX 2.

Almost but not quite.  I encourage the Posix smoothing and porting
efforts, but we aren't quite yet "Posix mature" to attract software
vendors who are unwilling to expend more than the "usual and customary"
efforts in porting.  MPE/iX remains above the "usual and customary"
effort level, although we are making great and encouraging strides.
Let's hope the trend continues (hey, if it works for NT...)

Jeff Kell <[log in to unmask]>
> >----------
> >From:  Ron Seybold[SMTP:[log in to unmask]]
> >Sent:  Tuesday, December 09, 1997 4:43 PM
> >To:    [log in to unmask]
> >Subject:       Re: [HP3000-L] A golden opportunity for MPE/iX
> <snip>
> >...NT doesn't have the 25 years of maturity and stability the HP 3000
> >counts on. He's counted on that stability, too. Nothing that's six years
> >old will have the durability of a solution that's performed more than four
> >times as long, surviving all fads while embracing all new technologies.
> <snip>