I've snipped much of the original message, so I hope you can wade through
the several threads interwoven under this general (off-) topic in the
embedded comments below.
Dick
-----Original Message-----
From: Philip.Belben_at_powertech.co.uk <Philip.Belben_at_powertech.co.uk>
To: Discussion re-collecting of classic computers
<classiccmp_at_u.washington.edu>
Date: Tuesday, July 06, 1999 11:57 AM
Subject: Re: OT: A call to arms (sort of)
>
>> Well, in my case it was supposed to lead to the development of a really
>> open hardware platform.
>>
>> As for Allison's comment that SPARC is to "high end" I have to disagree.
>> The SPARC architecture was initally a lot less complicated than the
PDP-11
>> architecture. It is the funky MMUs that get in the way.
>
>I'm reluctant to enter this rather heated debate, but:
>
>Has anyone tried to build a SPARC - compatible processor out of standard
SSI and
>MSI chips?
>
>Has anyone any thoughts on how easy, or otherwise, it would be?
>
>Philip.
>
>
>PS since I'm in here:
>
>Dick - it's all very well to choose a bus because it is a popular, de facto
>standard, but if you don't conform pretty closely to the standard, you may
as
>well not have a standard at all. And what other reasons are there for
choosing
>ISA?
>
>Tony - OPEN doesn't mean everything is documented. It means you can
connect
>what you like to it. And know what to do to get it to work.
>
>Chuck (wasn't it) suggested having an OPEN hardware architecture for which
>anyone can build periphereals and CPU cards which, if they conform to the
OPEN
>spec, will be guaranteed to work. I think this is an excellent idea.
>
Yes, and then someone jumped in with the notion that everyone who produces
hardware for such a bus should make every detail of his product available to
whoever wants it, including his competitors, and then, of course, decried
the use of custom logic even though the equivalent SSI/MSI version would
take a PCB a couple of acres in size . . . well, that's a bit of a hyperbole
. Would you believe square meters?
>
>However, there is nothing wrong with using undocumented cards on the
OPEN-spec
>bus which happen to work because they were designed for the system on which
the
>spec was based. Win-win! Those like you who want a fully documented
system can
>have one. Those like Dick who want to be able to use cheap cards from the
50p
>bin at that shop in Notting Hill can do that too. As I said, win-win.
>
When I suggested that it might be advantageous to be able to use boards
which are available from that 50p bin I wasn't thinking of myself. There
are people who have complained bitterly about the $2 cost of this and that,
and it's their interests I was considering. I use cheap stuff when it's
available, but I normally don't have to leave my house to find it. I've
actually never had a copy of the ISA spec, nor has it been a problem, though
I've designed 20+ boards in use even today for the ISA.
My position is that it's not necessary to have anything open. It's
generally straightforward to build something that works on the ISA as it is
for most any widely used bus. Since it's not in the interest of any
commercial manufacturer to build products as open as Tony Duell would like,
though it might have worked in the past (that's DISTANT past), it's unlikely
we'll see that happen. If you want a bus to use under public scrutiny, it
will be only a matter of time before someone finds a way to try to make bit
of money that way, and the "open-ness" will fade.
>
>P.
>
I'm not the one who is concerned (a) about whether "the bus" is open or (b)
about whether it's standard. I find it hard to buy cards which do what I
want yet don't adhere to some standard, however, so I buy stuff that's
commercially available, or build stuff that's compatible with what I have to
run with it.
>
>PPS I am concerned about the report that the open software movement is
losing
>momentum. Does anyone have any more news on this? I read quite recently
that
>Intel is trying to negotiate with some of the Linux community to get Linux
>available for their next generation of processors right from the start.
The
>difficulty seems to be NDAs, not surprisingly. I hope they solve it (e.g.
no
>disclosure until launch date, free thereafter).
>
>P.
>
The reason the open software is losing momentum is that the people who do
the work don't mind pitching in but do mind doing the work for people who
won't do their own work. These folks are finding that getting paid helps
their inclination to do useful work. OTOH, many of those who provided the
"momentum" though not willing to do the work, are not willing to pay others
for it either. As an example, look at LINUX, and then look at the state of
its documentation.
Dick
Received on Tue Jul 06 1999 - 13:58:36 BST