Hi, all!!!
Are there any old TCP/IP implementations like these to work on V7 or V6,
freely accessible in source form?
It could be very nice thing to learn from.
And possibly offtopic question - if I want ULTRIX (RISC) source license,
where should I get it from? (mostly interested in newest source) :)
All the best,
S.
Greg Lemis wondered,
> On page 182 of K&R 1st edition there's a reference to an
> implementation of C on the Honeywell 6000, with 9 bit bytes. There's
> no mention of whether it was running UNIX.
That one was a C implementation under GCOS. There
was another 9-bit one for the PDP-10 (not using Unix).
There was a 4x9 bit version of Unix for the Univac 1100
series, which ran Unix as a guest system over the
native EXEC OS.
The most exotic version was the BBN C-machine, which
had 20-bit words, 10-bit bytes.
Dennis
> Subject: Re: [TUHS] 6-bit, 7-bit and 9-bit byte UNIXes
> From: Norman Wilson <norman(a)nose.cs.utoronto.ca>
> To: tuhs(a)tuhs.org
> Date: Thu, 30 Sep 2004 20:47:22 -0400
>
> The very first UNIX ran on the PDP-7, which had 18-bit words.
>
> I happen to have the assembly-language source code to parts
> of that system. Many programs contain error-handling code
> that does something like this:
>
> lac d1
> sys write; 1f; 1
> jmp somewhere
>
> 1: 077012
>
> ...
>
> d1: 1
>
> Evidently the system thought in words in those days
> (the second argument to sys write is presumably a word
> count), but the single word written is a strong clue
> that 9-bit bytes were used, and that a certain concise
> error message that people love to complain about was
> there from the beginning (and why not?).
?
I would say "the PDP7 computer was word-addressable". In this
context, characters seem to have been packed as 9-bit half-words
in a big-endian fashion. No 'bytes'.
Maybe tomorrow I will be near my DEC literature archives, and see
if I can find some clues about PDP7 instructions that might deal
with half-words. If it's anything like the PDP8 of similar vintage,
there aren't any. Late in its life the PDP8 got a BSW "byte swap"
instruction to swap the half-words in the AC register. 6 bits,
of course.
carl
--
carl lowenstein marine physical lab u.c. san diego
clowenst(a)ucsd.edu
The very first UNIX ran on the PDP-7, which had 18-bit words.
I happen to have the assembly-language source code to parts
of that system. Many programs contain error-handling code
that does something like this:
lac d1
sys write; 1f; 1
jmp somewhere
1: 077012
...
d1: 1
Evidently the system thought in words in those days
(the second argument to sys write is presumably a word
count), but the single word written is a strong clue
that 9-bit bytes were used, and that a certain concise
error message that people love to complain about was
there from the beginning (and why not?).
Norman Wilson
Toronto ON
Natalia
I don't know of any non-8-bit Unix systems, but Multics, on the GE645 at
least, had a 36-bit word. See
http://en.wikipedia.org/wiki/Multics
James
----- Original Message -----
From: "Natalia Portillo" <iosglpgc(a)teleline.es>
Date: Thu, 30 Sep 2004 20:07:46 +0100
To: <tuhs(a)tuhs.org>
Subject: [TUHS] 6-bit, 7-bit and 9-bit byte UNIXes
> Hi!
>
> Was there any UNIX with 6-bit wide, 7-bit wide or 9-bit wide bytes or all
> UNIXes are 8-bit wide bytes?
>
> Regards
>
> _______________________________________________
> TUHS mailing list
> TUHS(a)minnie.tuhs.org
> http://minnie.tuhs.org/mailman/listinfo/tuhs
Hi,
I have a copy of the WEBSTER server and client port to UNIX by
* David A. Curry
* Purdue University
* Engineering Computer Network
* April, 1986
I see in the doc directory a very TOPS-20-ish docuement webster.hlp which
describes the way to invoke the client:
@WEBSTER word-to-define
I'm assuming the @ is the TOPS-20 prompt. It also says you can use
If you want to look up more than one word, just do
@WEBSTER<return>
and you will be prompted with
Word:
Type the word, or hit <return> to exit.
But if the @ prompt wasn't enough evidence of TOPS-20 you also get:
<escape> and "?" are used the same way in Webster as in most programs.
<escape> tries to complete what you have typed so far, and "?" lists
those words that match your partial word.
Which is pure TOPS-20 "COMND JSYS". See this page for what "COMND JSYS" is:
http://pdp-10.trailing-edge.com/decuslib20-01/01/decus/20-0002/comnd.doc.ht…
There is some cryptic mentioning about EBCDIC conversion as if the dictionary
data went from ASCII to EBCDIC and back to ASCII causing some degradation.
ebcdictp.ememo talks about the format of the EBCDIC tape (not that I have one)
and errors.ememo says:
2) Pronunciation records.
Three classes of errors occur here.
During the translation to EBCDIC occurances of the glyphs *( and
)* in the pronunciation records were treated in the same way as in other
records, that is encoded as <( and >) (representing left and right
braces), rather than left as is. Thus a schwa precceding or following an
optional phoneme was lost.
I just blew a few minutes looking it over today and ported it to FreeBSD/AMD64
catching a
char word[BUFSIZ];
isnumber(word)
bug and some other minor things.
Does anyone else have this treasure running? I like it better than dict for
etymologies.
Hi,
I have a copy of the WEBSTER server and client port to UNIX by
* David A. Curry
* Purdue University
* Engineering Computer Network
* April, 1986
I see in the doc directory a very TOPS-20-ish docuement webster.hlp which
describes the way to invoke the client:
@WEBSTER word-to-define
I'm assuming the @ is the TOPS-20 prompt. It also says you can use
If you want to look up more than one word, just do
@WEBSTER<return>
and you will be prompted with
Word:
Type the word, or hit <return> to exit.
But if the @ prompt wasn't enough evidence of TOPS-20 you also get:
<escape> and "?" are used the same way in Webster as in most programs.
<escape> tries to complete what you have typed so far, and "?" lists
those words that match your partial word.
Which is pure TOPS-20 "COMND JSYS". See this page for what "COMND JSYS" is:
http://pdp-10.trailing-edge.com/decuslib20-01/01/decus/20-0002/comnd.doc.ht…
There is some cryptic mentioning about EBCDIC conversion as if the dictionary
data went from ASCII to EBCDIC and back to ASCII causing some degradation.
ebcdictp.ememo talks about the format of the EBCDIC tape (not that I have one)
and errors.ememo says:
2) Pronunciation records.
Three classes of errors occur here.
During the translation to EBCDIC occurances of the glyphs *( and
)* in the pronunciation records were treated in the same way as in other
records, that is encoded as <( and >) (representing left and right
braces), rather than left as is. Thus a schwa precceding or following an
optional phoneme was lost.
I just blew a few minutes looking it over today and ported it to FreeBSD/AMD64
catching a
char word[BUFSIZ];
isnumber(word)
bug and some other minor things.
Does anyone else have this treasure running? I like it better than dict for
etymologies.
