24 Jun
2007
24 Jun
'07
noon
Dave McGuire <mcguire at neurotica.com> wrote:
On Jun 23, 2007, at 9:05 PM, Michael B. Brutman
wrote:
Even the current PPC hardware is not 'standard' .. there are some
extensions to support the single level store. 'Tagged
pointers' (ie: pointers marked as trusted by the OS) are the single
biggest oddity that I can think of, and that requires hardware
support.
Sadly, the AS/400 line (iSeries) is slowly dying off ... IBM can't
market it's way out of a paper bag, and the 'cash cow' status the
machine has had has caused it to fall too far behind.
Last year I found myself digging around in the job market. As
most people know, there's just not much technical work in Florida
compared to other areas, so I was running scared for a while. One
thing I noticed is that all the AS/400-related jobs in the whole damn
world seem to be here. I figured that was because of their extensive
use in the health care industry, and while Florida's reputation for
being filled to overflowing with retirees is inaccurate, it's not
entirely unfounded.
I wonder how quick those types of places will be to abandon their
AS/400 gear once IBM EOLs it.
Well, almost all of them have probably been migrating the same code
Their software is pretty unusual, but there's
nothing that weird
about
the hardware, is there?
Not anymore, with the change to PowerAS some years back.
The original
AS/400s were more odd in the hardware sense.
The whole concept of single level storage is amazing in that works
at all. from hardware to hardware for two decades already now.
I have
some appreciation for how the abstractions were done to make this
possible. But:
Literally, you could drop an AS/400 into my basement, along with all the
software and all the manuals, and I would find myself unable to use it.
I am that completely incapable of working in that architecture or
(even worse) corporate mindset. Obviously lots of people do useful
work in it all the time, but my quote-highly-trained-mind-unquote
is simply unable to adjust to that mindset.
I have written useful assembly code for 1802's, where any register
can be the program counter. I have microcoded for AMD bit-slice.
I have built my own logic devices out of vacuum tubes and relays.
I have rewritten hairy device driver modules for hardware with bizarre
undocumented side-effects on real-time OS's running $10Billion peripherals.
I eventually figured out that to shut down a Windows box, you click on Start
(that was a very tough hurdle for me - you'd be surprised!)
Yet I cannot, and probably will never be able to, use an AS/400
application succesfully. The stumbling blocks for me are really
that enormous.
Tim.