|
|
 Up |
|
|
  |
Author: JKJK Date: Nov 26, 2007 14:37
Hi folks,
This is a simple request for historical information. No, I
don't want to re-implement Unix in Forth. I am merely curious
about one thing:
When they were working on the first couple of iterations
of Unix, in the early 1970's, were Ken Thompson and Dennis
Ritchie *aware* of Forth and/or Charles Moore? (And, was Moore
aware of activity related to Unix, at the same time? [OK,
two things.])
Any anecdotes, musings, vague handwaving, hazy memories,
or links to relevant historical data would be welcome. The
almighty Google is not of much use, except to imply (by absence
of relevant data) that there seems to have been very little
cross-pollination between the Forth and Unix phyla of the family
tree of computing. I may post this query to c.os.unix and c.l.c
as well, but since I'm an irregular regular on c.l.forth I thought
I'd try here first.
Thanks,
-- Joe Knapka
|
| |
|
| | 91 Comments |
|
|
Author: JKJK Date: Nov 26, 2007 22:42
Elizabeth D Rather wrote:
> JK wrote:
>> Hi folks,
>>
>> This is a simple request for historical information. No, I
>> don't want to re-implement Unix in Forth. I am merely curious
>> about one thing:
>>
>> When they were working on the first couple of iterations
>> of Unix, in the early 1970's, were Ken Thompson and Dennis
>> Ritchie *aware* of Forth and/or Charles Moore? (And, was Moore
>> aware of activity related to Unix, at the same time? [OK,
>> two things.])
>
> To the best of my recollection, the answers are "no" and "vaguely".
Thank you, Elizabeth.
|
| Show full article (2.33Kb) |
|
| | no comments |
|
|
Author: Elizabeth D RatherElizabeth D Rather Date: Nov 27, 2007 00:23
JK wrote:
> Elizabeth D Rather wrote:
>> JK wrote:
>>> Hi folks,
>>>
>>> This is a simple request for historical information. No, I
>>> don't want to re-implement Unix in Forth. I am merely curious
>>> about one thing:
>>>
>>> When they were working on the first couple of iterations
>>> of Unix, in the early 1970's, were Ken Thompson and Dennis
>>> Ritchie *aware* of Forth and/or Charles Moore? (And, was Moore
>>> aware of activity related to Unix, at the same time? [OK,
>>> two things.])
>>
>> To the best of my recollection, the answers are "no" and "vaguely".
>
> Thank you, Elizabeth.
>
>> Chuck and I were aware that Unix was in development, and we actually ...
|
| Show full article (4.01Kb) |
| no comments |
|
|
Author: Stan BarrStan Barr Date: Nov 27, 2007 03:07
On Mon, 26 Nov 2007 15:52:58 -1000, Elizabeth D Rather forth.com>
wrote:>
>Several CP/M and DOS Forths were developed in the early 80's. FORTH,
>Inc. developed a Forth for VAX/VMS in the mid 80's. The first I heard
>of Unix Forth development was ~1984-5. In the 70's, all Forths were
>dedicated, bare-metal, standalone systems.
The FIG Forth I run on my PDP-11, under RT-11, is copyright January 1980,
so was presumably running in the late '70s, but that's just me being picky!
I also run 6th Edition Unix (1975) and I've not heard of a Forth for it,
but if anyone knows of one, or for 7th Ed. Unix (1979) I'd love to hear
about it. It would be easy enough to write one, but that would be
cheating.
--
Cheers,
Stan Barr stanb .at. dial .dot. pipex .dot. com
(Remove any digits from the addresses when mailing me.)
The future was never like this!
|
| |
| no comments |
|
|
Author: Stephen J. BevanStephen J. Bevan Date: Nov 27, 2007 07:10
Elizabeth D Rather forth.com> writes:> Incidentally, I think the low-end machine was a PDP-8, not PDP-7
> (which I never heard of).
|
| |
| no comments |
|
|
Author: Stan BarrStan Barr Date: Nov 27, 2007 10:28
On Mon, 26 Nov 2007 22:23:19 -1000, Elizabeth D Rather forth.com>
wrote:
>Incidentally, I think the low-end machine was a PDP-8, not PDP-7 (which
>I never heard of).
The earliest Unix (1969-70) ran on PDP-7 and -9.
The second ran on a -20 and the third and subsequent versions on various
-11s.
(Source: "The UNIX Time-Sharing System", Ritchie and Thompson, 1974)
I've got the source for a modern-ish (1984) PDP-8 Forth, by John Wilson,
for TSS-8, but it looks a bit messy, and I've not got it to run yet.
The trouble is I know very little about TSS-8, the operating system.
--
Cheers,
Stan Barr stanb .at. dial .dot. pipex .dot. com
(Remove any digits from the addresses when mailing me.)
The future was never like this!
|
| |
| no comments |
|
|
Author: JKJK Date: Nov 27, 2007 22:01
Stephen J. Bevan wrote:
> A more logical choice would have been BCPL. However, given the route
> of BCPL->B->C we can see that K&R moved from a language with little or
> no static typing and an interpreted implementation towards one with
> explicit types and a native code compiler. That doesn't rule out
> Forth but it doesn't make Forth any more an obvious candidate than
> say Lisp.
At the time, I suppose not. The Lisps of the day were certainly
a lot simpler than Common Lisp.
-- JK
|
| |
| no comments |
|
|
Author: Anton ErtlAnton Ertl Date: Nov 28, 2007 03:14
>JK kneuro.spam.net> writes:>> It occurred
>> to me (as I'm sure it has to many others) that if the Unix
>> guys had been comfortable with Forth, it might've been an
>> an obvious choice for a small, portable, high-level language
>> in which to re-implement Unix (which initially was written
>> in PDP-[7|11] assembly).
...
>A more logical choice would have been BCPL. However, given the route
>of BCPL->B->C we can see that K&R moved from a language with little or
>no static typing and an interpreted implementation towards one with
>explicit types and a native code compiler.
|
| Show full article (1.44Kb) |
| no comments |
|
|
Author: Stephen J. BevanStephen J. Bevan Date: Nov 28, 2007 07:00
> I am pretty sure they wanted a native code compiler for speed, but
> Forth can be compiled to native code. Concerning the types, my
> impression (probably from the HOPL-II paper on C) is that they added
> types because they were running on a byte-addressed machine and
> therefore now needed more types than just the machine word that BCPL
> worked on, and which was sufficient for word-addressed machines.
> Forth has solved this problem, too, although differently.
Indeed and it would be easy to extend BCPL using a similar approach.
However K&R chose to go with explicit types which arguably is a more
complicated solution to implement. From this I infer that they
implicitly or explicitly rejected the simpler approach in favour of
static types. Perhaps, like John Doty, they fundamentally don't like
having a multiplicity of words to access memory.
> So I don't think these two issues would have excluded Forth.
I agree which is why, in the one line of my message you didn't quote,
I wrote :-
>> That doesn't rule out Forth but it doesn't make Forth any more an
>> obvious candidate than any Lisp.
|
| |
| no comments |
|
|
|
|
|
Author: Andy ValenciaAndy Valencia Date: Nov 28, 2007 07:46
Anton Ertl mips.complang.tuwien.ac.at> wrote:> ... Concerning the types, my
> impression (probably from the HOPL-II paper on C) is that they added
> types because they were running on a byte-addressed machine and
> therefore now needed more types than just the machine word that BCPL
> worked on, and which was sufficient for word-addressed machines.
I remember an interview--I believe it was Dennis Ritchie--where he commented
on how much of a win it was when the compiler first started understanding
types. Before that, all pointer arithmetic was manual, with explicit
additions of size and everything. After, "ptr+1" would be "ptr" as advanced
to point to the next object of its (pointed to) type. They apparently found
that this permitted vast improvements in source code compactness and
readability.
Regards,
Andy Valencia
|
| |
| no comments |
|
RELATED THREADS |
|
|
|
|
|
|
