Re: Compiling with another compiler than gcc - OT reply

On Tue, Sep 18, 2007 at 09:34 ,
freebsd-current-request_at_freebsd.org exclaimed "Las Cucarachas
entran, Pero no pueden salir", and then rambled on saying with:
 

> Date: Mon, 17 Sep 2007 23:30:06 +0200
> From: Hartmut Brandt <hartmut.brandt_at_dlr.de>
> Subject: Re: Compiling with another compiler than gcc.
> To: Wilko Bulte <wb_at_freebie.xs4all.nl>
> Cc: freebsd-current_at_freebsd.org, Anders Magnusson <ragge_at_ludd.ltu.se>

> Wilko Bulte wrote:
> > On Mon, Sep 17, 2007 at 08:38:35PM +0200, Hartmut Brandt wrote..
> >> Anders Magnusson wrote:

[mucho deletia - wjv]

> I've wrote an emulator for the KDJ11A (11/73 I think) more than 10 years 
> ago before I gave my machine to Grog. It runs RSX11M[+], RT11, v[567], 
> 2.11BSD at light speed :-) It's on people.freebsd.org/~harti. Compiles 
> 2.11BSD in less than an hour. It's probably easier than to put up your 
> 11/73 :-)

> About pcc: I've a z8000 based computer with a Unix System III on which 
> uses pcc as the system compiler. It's so buggy that I wonder how the 
> system even runs. But I suppose these bugs are mostly in the z8000 
> backend (no sources to check, though).

pcc rang a bell.  And was the Z8000 based computer with Unix
Sysstem III running on it and Onyx.

I had to do so many things to get things to compile correctly.

For some reason it or the pre-processor like to sign extend
commands even when parity was off, as it would set the parity
bit and sign extend 8 bit commands to 16 bit - so instead
of returning the correct info from the device which I was querying
I would get all sorts of errors on SOME codes I sent.

I found the sign extend by noting than only every other character
in the ASCII set returned with an error.  I got that fixed
by masking the data sent or returned - I forget which - and I never
had fond memories of that machine.

There was no serial communications built it, so I took
a Radio Shack Model 4 portable - with an early version of
Kermit - running about 40K compiled - and set the transmission
to pause 1 second between each line transmitted.

That was because the only way to get things in was to use
the 4 as a terminal, bring up vi, transfer at the slow rate,
exit vi, and them compile kermit.

THe next step was to transfer hundreds of update COBOL code into
the machine.  The only input to that was a 40MB tape and I think
when I worked on this the machine was no longer supported.

Bill

-- 
Bill Vermillion - bv _at_ wjv . com