On May 13, 2013, at 7:45 AM, Holger Veit wrote:
Not having seen the PPC source, but the
86/286/386 monster of OS/2-kernel source (3.x), I can only agree. I consider it almost
impossible, without a complete rewrite and then only - as in the PPC implementation - with
largely different APIs to port this thing to anything else, including PDP-11. Given the
386 ring architecture, it would be more likely to transfer a look-alike to a VAX, but not
to a PDP with MMU. But even then; the question is: what benefit would there be? DEC
systems were already a dying platform at the time of OS/2, and lack of any compatibility
to the original x86-based system would gain nothing. Even the PPC was not a success, not
only because of the obscure CHRP architecture no other vendor used, but also because of
the lack of killer application or migration path. Apple those days did it better in
providing an emulator to run older 68K apps on the PPC.
I dunno, I think the PPC
was a success, just not as much on the desktop.
There are LOADS of PowerPCs running high-end embedded stuff in cars and
telecom systems, and all three of the previous generation of game
consoles used a PowerPC of some sort as the CPU.
I was aiming at desktops, and
there in particular at its role as a
potential successor of the x86 platform. It failed there, mainly because
of the inability to come to an accepted (or acceptable) standard. The
same happened to MIPS at that time (WinNT on MIPS - no applications).
This is to a high degree due to the failure of Microsoft to understand
the impact of software availability - they are repeating it nowadays
with Win8 RT, in the markets where IOS and Android are already
ubiquituous. Apple, as mentioned was an exception because they provided
a migration path.
The embedded area is an entirely different battlefield - products don't
need and maybe producers don't want any compatibility at all. I agree,
the PPC surely had its application area there although it is meanwhile
going to be replaced by less power hungry ARM families.
--
Holger