Re: Market pressures for more reliable software

["Gary Labowitz" <garyl@enter.net>]

I'm curious what you mean by "relocation" that you were shocked that S/360
didn't have it. From the very outset the OS/360 design included a relocating
loader, which was very advanced for its time. The hardware redirection of
memory addresses (dynamic address translation -or DAT) came later (as I
recall about four years).
The development activity in Poughkeepsie was very untypical at the time.
The MP65 was pretty much a late-comer to market with multiprocessing, but it
never really went anywhere. Neither, BTW, did Burroughs, NCR, or Bendix.
Univac held on much longer. I remember losing a large contract to Univac in
the early '70's, but they later switched back to IBM.
As I went about learning PC's (and Apple) operating systems (including
Win32) I was struck with the fact that much of what was the "new" thing was
just a reimplementation of what we did on the S/360 in the mid '60's. Names
changed, and there was a lot more packed into the hardware, but a great deal
was the same. It has drifted a bit,by which I mean that there are some new
concepts pushing PC development in a different direction from the way OS/360
did things, but we were very close to what OOP is all about. IBM's biggest
mistake (like Apple) was trying to keep everything proprietary for market
advantage, and losing to the clones. I think Java is finally breaking into
the same sort of hold Microsoft has used by its "fully integrated" approach.
I will, of course, learn C#, but frankly I'm having more fun now with Java
than with VB (I never committed to C/C++). Perhaps it's just that it's newer
to me, but as long as the performance improves steadily, Java may well be
the challenge Microsoft needs for change, not court mandated crippling.
Gary
"Shmuel (Seymour J.) Metz" <nospam@acm.org> wrote in message
news:3B387768.FF5E9744@acm.org...
>
>
> Ken Garlington wrote:
>
>
> > Let's say, just hypothetically, that you've been programming since 1960.
How
> > long can you remember trends in the computing industry (which,
presumably,
> > is different from either academic computing or hobbyist programming)?
>
> Trends were also different between scientific and commercial computing.
> Centralized development was much more common in the commercial sector.
>
> > In fact, let's go further - does your memory confirm that the OS/360
> > development effort described in Brooks was fairly typical (if not
advanced)
> > for industrial computing in the early-to-mid 1960's?
>
> I remember being shocked that the S/360 didn't include any relocation
> capability. I remember that OS/360 inherited some nice features from
> IBSYS/IBJOB. Overall, my memory confirms that it was less advanced but
more
> comprehensive than what was available from other vendors. In particular,
both
> UNIVAC and Burroughs had working multiprocessors on standard lines well
before
> the 65MP or even the 9020.
>
> --
> Shmuel (Seymour J.) Metz
>
>