Re: Apple PowerMac G5
Em Sat, 05 Jul 2003 18:36:51 +0800, debia escreveu:
> The S/360 was open enough that others could clone it, and they did.
Wrong. There was no commitment by IBM on documenting or preserving
interfaces, so clone-makers and plugin compatible module vendors were left
to reverse-engineering every new change -- and IBM was known for changing
interfaces just to lock out third-party stuff.
> Mostly, though, you still got to run IBM operating systems on them.
Which were only available because the US government forced IBM to sell
> The S/370 was open enough that others could clone it, and they
> did. In some cases, vendors (such as Fujitsu and Hitachi) provided
> their own operating systems. However, they were derived illegally
> from IBM's OS/VS family - there were court cases and settlements to
> prove it. They _could_ have written their own operating systems, and
> may have done so for the Japanese market.
So how open that makes it? Not at all, in my book. EBCDIC, changing
undocumented interfaces... that's not my definition of openness.
> The PC familiy was open enough that others could clone it, and they did.
This is a totally different story from the S/360 architecture. Because
(1) the IBM PC was low priority and (2) IBM was a later entrant at a
market then already perceived as saturated by Apple, TRS-80, Sinclair and
others, IBM was willing to simply assemble a system based on available
components, contrary to their first entry on the microcomputer market,
which was a totally closed system no one paid attention. So this wasn't
open by design, but by accident.
> The impendiment to writing your own OS for the S/360 was not the
> available information - back then the OS was free, and I think came
> with source.
Bundled, not free. The source was available but controlled.
> The impendiment to writing your own OS for the PC was not the
> available information, but rather the availability of tools and the
> number of people using the PC. As those problems were resolved,
> people wrote the tools and OS operating systems, mostly in the Unix
> mould, have proliferated.
I am not talking simply writing OSs here, but writing open OSs
on an open platform. The PC platform allowed proprietarisation by MS
because IBM tried to make it even closer with IBM OS/2 and PS/2, thus
allowing MS and the clone-makers to be seen as relatively open, with
EISA, MS-DOS, MS OS/2 and later MS Windows. Later there was the ACE
consortium, which tried to create an open platform but was stabbed on
the back by MS and Intel when they saw they could get away with it,
because (1) their products had became good enough to prevent wholesale
migration to open systems and (2) the open systems camp was seen to be
in disarray because of the desktop wars (OpenLook vs CDE, X vs NeWS,
everyone vs NeXTStep), the Unix wars (System V vs BSD, X/Open vs POSIX
vs Sun) so on.
> As for Compaq killing Alpha and PA-RISC, that was driven by its
> perception of how best to make a dollar. NT 4 was available for
> IA32, Alpha, PPC and HP-RISC.
There was never MS WNT for the PA-RISC. There was for SPARC,
but never saw the light of the day. Likewise Solaris for the PowerPC
was dead at arrival.
Moral: the Unices were not truly open, their RISC platforms
were never truly open, and this opened the doors for completely or
almost completely proprietary systems, that is, Wintel.
Just a more recent example: FireWire vs USB. FireWire is
clearly the superior standard technically, and it is an IEEE
standard. Nevertheless patents and dumb marketing tactics by Apple
(refraining from implementing FireWire at initial iMacs and iBooks,
charging too much, forbidding everyone else from using the FireWire
name) left us with USB everywhere, even where it is totally inadequate
such as external storage, scanners and colour printers.
The lesson of history is that a quasi-open standard dumbfully
marketed, even if superior, will loose to an inferior proprietary
interface skillfully marketed by a monopolist.
> When MS withdrew support for Alpha, that pretty much killed the
> Alpha as far as I can see.
Yes, but why is that? Because NT was optimised only for the
IA32, where it inherited MS-DOS and OS/2 assembly, time-proven code,
agains all-new C-only RISC code in the other platforms. Not to
mention ridiculous availability of workstation and other applications
/ \ Leandro Guimarães Faria Corsetti Dutra +41 (21) 648 11 34
\ / http://br.geocities.com./lgcdutra/ +41 (78) 778 11 34
/ \ Responda à lista, não a mim diretamente! +55 (11) 5686 2219
Dê-me nota se te ajudei: http://svcs.affero.net/rm.php?r=leandro