[rescue] AT&T 3b1 Starlan software

[Jonathan C. Patschke]

On Thu, 13 Feb 2003, Francisco Javier Mesa-Martinez wrote:

> I am sorry but if you compare the IBM PC to today's architectures, of
> course it is not competitive. But as usual you have to see the PC when it
> was first introduced, it was actually a pretty good machine.

Bullshit.  It was broken-as-designed.  It was a design requirement that
it be suitably stripped-down so as not to steal the thunder from IBM's
bigger iron.  This was instrumental in their choice of the Intel 8088
over the more capable Motorola 68000.

> Standard well documented bus for expansion,

Standard if you were IBM.  They reused it from some earlier (failed)
project.  I forget the name of it, but I could probably look it up.
This well-documented bus didn't do any sort of device negotiation.  If
you plugged a device in slot 4, you -still- had to configure it.  The
system wasn't smart enough to say "hey, you're in slot 4.  You get these
resources".  Every system back then that didn't just have a dumb
backplane for an interconnect could do that.  As a result, PC video
cards -to- -this- -day- are expected to map their initialization code to
a particular memory address and attach to a particular IRQ line.  Thus,
cards that -could- work on any platform (due to buses like PCI and AGP
which are available all over the place) can only work on PCs because
someone at IBM didn't think ahead.

The bus may have been okay (even if it was designed around the processor
instead of the system as a whole), but the system interface to that bus
was and is -horrible-.

> moudular design,

This was unique?  The Apple ][ had this.  Hell, my TI 99/4a had this!
Even the MITS Altair had this!!

> scalable storage... what a concept!

Yeah, you could choose one or two single-sided floppy drives.  Woo!  You
didn't get a choice of fixed discs until DOS 2.0, IIRC, and I still
don't know if the original PC was capable of -booting- from a hard drive
without a BIOS replacement.

> It had its faults, but then there are no perfect systems anyway. But for
> that time, the PC was a darn good design!

Which is why the PC had the "640k barrier" and the "IRQ cascade"
nightmare that haunts the IA32 platform to this very day.  These were
all things that could've been avoided with then-current technology, if
they thought the platform would've been anything more than a toy.

-- 
Jonathan Patschke  *)  "It's not about who's right and who's wrong...
Thorndale, TX      (*   it's about who works for the government and who
                   *)   doesn't."                        --Dave McGuire