[ncolug] Re: Minix3
- From: Chuck Stickelman <cstickelman@xxxxxxxxxx>
- To: ncolug@xxxxxxxxxxxxx
- Date: Sat, 12 Nov 2005 23:18:32 -0500
Henry Keultjes wrote:
Larry:
The difference between the old Pick and the Pick I am trying to
reinvent is that my version can take Open Source drivers for any old
and new hardware and adapt those drivers to that reinvented Pick
fairly easy. My idea is not reinventing the wheel but, as a matter
of speaking, just taking the tire off and putting that bold lettering
and whitewalls on the inside.
My plan has always been to only run on PowerPC and to only support a
very limited selection of hardware like only Seagate HDD's and only
AOpen peripharels like CD-Rom and DVD's. However, because this is all
Open Source, the choices to everyone else are never limited by my
choices Anyone who wants to port this to any other processor can do
so and anyone who wants to adapt any other hardware to this platform
can do so.
In contrast, it was the proprietary aspect of Pick that was the
problem of not having choices. Noone but Pick Systems had the legal
right to the source code that facilitated making those choices.
Under Open Source we can all have it our own way. Obviously it might
not be economical nor practical for everyone to adapt drivers to that
platform. But that's like democracy. If you don't have the votes you
need to live with how the majority rules or you become a rebel like
Henry.
Henry
larry wrote:
Henry Keultjes wrote:
larry wrote:
I think what you are trying to say is the following:
"I want an operating environment, where I can play around with
stuff like I/O, security, and screen formatting, just by editing
some text files."
"Furthermore, I would like to be able to make various hardware, no
matter how old, work in this environment, again simply by editing
some text files."
"Finally, I would like to customize the applications that run in
this environment, by (you guessed it) editing some text files.
Does that sum it up, Henry?
Indeed!
Henry
Good, I am glad we established this.
The reason you could "play around with screen formatting" back then
was that the screen was an 80x25 matrix of alphabetic characters.
(This was also back in the days when running an application meant
that you needed to know how to READ - but I am not going to go down
that road right now.)
The reason I/O was easily configurable via text files in Pick was
because of the limited hardware support. If you look at what hardware
was actually supported on the incarnation of Pick that ran on an NCR
Tower, for example, you could practically name the items on one hand
- maybe 2 or 3 models of tape drive, a couple disk drives, that was
pretty much it.
The part of my original message where I said, "...various hardware,
no matter how old," was thrown in simply because I know that's what
you would like. Yet, it is the only part that was never true for
Pick, or any other OS. Of course, Linux is about 500% more likely to
support any given "old hardware" compared to any other OS, but even
that has it's limits.
Releases of Pick were frozen in time just like any other OS, and a
given release supported whatever hardware the company who ported it
wanted it to. Your big problem, as I see it, is that there are no
qualified engineers (working for large hardware vendors or not,) who
have any incentive to do a port to modern hardware. Why not?
The hardware industry has changed since the 80's. The steady price
erosion of computing hardware in general means, basically, that if
you want to make any money whatsoever on computer hardware, you had
better be Chinese. (Taiwanese, to be exact.) This led to
standardization and commoditization of all but the largest computing
hardware.
Next, (we are in the 90's now) we had standardization and
commoditization of the OS, which, as many have said here, is no more
than "an extension of the hardware" to most users.
So, back to the present, we now have an array of 2 or 3 low cost (or
no cost) operating environments to choose from, compared to the 80's.
No one really cares what hardware they need to run on, as long as
they can get the whole solution for less $ than last year.
No one wants to port an old OS to modern hardware, because Linux and
Windows are available, supported, and cheap. Engineers are very
practical people.
You may say, "...but today's server OS's aren't cheap!" Oh yes they
are. A copy of Pick, when it was popular, went for several thousand
(1980's) dollars. When they later replaced it with Unix, the price
was only a little less. And don't forget, the hardware was MORE than
the OS. The typical Tower solution that was still running all
Phar-Mor pharmacies (are they still in business?) only ten years ago,
was around $25,000. That was for a 68040 processor running Unix to
drive a dozen or so dumb terminals.
So my point on hardware is, "you have to go with the flow." You
aren't going to see text-file configuration of hardware any more,
because hardware is way more complex, the options are endless, and
the manufacturers of hardware are spread all over Asia. Not to
mention that there's just a lot more of it to choose from in the
first place. This dictates that we are still at the mercy of hardware
manufacturers (just different ones.)
I do have some good news, though... the "application customization
via text files" part - that is alive and well, and oddly enough, is
the subject of my presentation next month.
I understand writing a new platform to only support on CPU architecture,
that's the way most of them start.
Though I don't understand only one or a few HD's - ATA drives have a
standardized interface. The whole point of an interface.
While I admire your goal, I have to agree with Mike K. about asking the
question "How does this have value to the LUG?" it seems like a
long-shot to me...
Taking drivers from one platform and using them on another is only easy
if both platforms have the same API. The more alike the two platforms
are the easier. Since PICK in any incarnation is so very different from
Linux or the BSDs it's going to be quite a bit of work to get things
ported. Seems like the two choices are to get A LOT of people excited
about the project so they'll donate their time and effort. (Which
unfortunately isn't the folks in this LUG -- we don't have any seasoned
driver developers...) or to pay a bunch of seasoned driver developers to
write the code you need.
Personally, I've heard a lot of talk about PICK, I know that it's a lot
like a DB that I'm familiar with (AppGen), but don't see anything that
makes it all that great. I've heard a lot of talk about RISC CPUs, but
haven't heard a single thing about what they can do that CISC can't do.
Same goes with the whole micro-kernel vs. monolithic thing. Find me one
application that can only be implemented on PICK using a micro kernel
based OS on RISC CPUs, and I'll start to listen. Otherwise I have only
a limit amount of attention I can expend on the topic and I've chosen to
make that focused on Linux.
Chuck
begin:vcard
fn:Chuck Stickelman
n:Stickelman;Chuck
adr:;;3068 Noblet Rd.;Mansfield;Ohio;44903-8634;USA
email;internet:cstickelman@xxxxxxxxxx
tel;home:419-884-8240
tel;cell:419-610-6638
x-mozilla-html:TRUE
version:2.1
end:vcard
Other related posts:
