Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

seeking a special Forth

3,088 views
Skip to first unread message

DavidM

unread,
Jun 28, 2008, 8:23:55 AM6/28/08
to
Hi,

Does anyone know of a Forth implementation that:

- is written almost entirely in forth, save for the absolutely
smallest possible number of primitives
- has its handful of primitives written in highly portable C:
- can compile for 32-bit or 64-bit, little or big-endian
- is ANS compliant
- has a good C API, easy to add a small number of extra C
primitives

I've been working with FICL - it has around 12k lines of C - I'd be
curious to work with a Forth with only a few hundred (or even a few
dozen) lines of C.

Thanks in advance

Dave

sp...@controlq.com

unread,
Jun 28, 2008, 6:09:58 PM6/28/08
to

Having implemented a couple of Forth's in C, it begs the question, just
what is the minimal wordset with which you can bootstrap a Forth
implementation? Is there any literature on this?


---- Posted via Pronews.com - Premium Corporate Usenet News Provider ----
http://www.pronews.com offers corporate packages that have access to 100,000+ newsgroups

foxchip

unread,
Jun 28, 2008, 7:03:59 PM6/28/08
to
On Jun 28, 5:23 am, DavidM <nos...@nowhere.com> wrote:
> Does anyone know of a Forth implementation that:
>
> - is written almost entirely in forth, save for the absolutely
> smallest possible number of primitives

Minimal ANS Forth and eForth have a small number of primitives.
But there is a tradeoff with performance that drops off rather
steeply when the number of primitives drops below the minimal
practical number.

It is a little like asking what is the minimal instruction set.
It is an academic question not a practical one. In reality no
one uses the absolute minimum. eForth is more mature and more
complete than MAF and has a few more primitives. About 80% of
code can easily consist of nothing but about 30 primitives and
when you drop below that code size will go up considerably and
performance will drop considerably.

> - has its handful of primitives written in highly portable C:

There is a version of eForth with primitives in C. You can
combine those primitives with any other version of eForth you
choose.

> - can compile for 32-bit or 64-bit, little or big-endian

eForth has 16/32/64 bit versions with little more than a
declaration, and handles endian issues.

> - is ANS compliant

Both MAF and eForth are standard.

> - has a good C API, easy to add a small number of extra C
> primitives

The C version of eForth should meet those needs and can be
mixed with later versions of the high level code easily.

> I've been working with FICL - it has around 12k lines of C - I'd be
> curious to work with a Forth with only a few hundred (or even a few
> dozen) lines of C.

eForth typically has only a couple of hundred lines of assembler for
the primitives. I haven't counted the lines of C code for primitives.

Most people assemble the kernel and then compile the rest from high
level Forth. Some people convert the whole thing to assembler and
some convert the kernel to C. Some people metacompile the whole
thing from Forth.

The eForth 1.x series was DTC or IDTC while the 2.x through 5.x and
beyond may be STC or STC with native code inlining. A peephole
optimizer
can be added easily to most any version. There are also several
porting
manuals for eForth and porting is pretty easy.

> Thanks in advance
>
> Dave

Best Wishes

Robert Miller

unread,
Jun 28, 2008, 11:08:58 PM6/28/08
to

<sp...@controlq.com> wrote in message
news:alpine.BSF.1.10.0...@saddam.controlq.com...

>
> Having implemented a couple of Forth's in C, it begs the question, just
> what is the minimal wordset with which you can bootstrap a Forth
> implementation? Is there any literature on this?
>

I don't know how relevant this might be to modern implementations, but about
thirty years ago the answer to that question was IIRC sixteen. We had been
looking at Forth for a 6809 based test system and our software guru (not me)
found a paper somewhere that identified the minimum wordset from which the
whole language could be built. It took him only a week to get a limited
version of FORTH up and running. MVP Forth was the model on which the
implementation was expanded. It was only another couple of weeks until the
whole language was implemented. Frequently used words were later recoded in
assembly language.

Bob


Oyster

unread,
Jun 29, 2008, 4:13:39 AM6/29/08
to

The Beez'

unread,
Jun 29, 2008, 4:15:40 AM6/29/08
to
On 28 jun, 14:23, DavidM <nos...@nowhere.com> wrote:
> I've been working with FICL - it has around 12k lines of C - I'd be
> curious to work with a Forth with only a few hundred (or even a few
> dozen) lines of C.
Well, 4tH is about 5.5K lines of code, but it was written as a
library, so take what you need. Furthermore, it was designed to be
highly portable and easily interface with C. Whether 16, 32 or 64 bit
- 4tH won't care.

It comes with a 100 page development guide, so changing it shouldn't
be a problem. There is even a mechanism to take your personal
enhancements from one version to the next.

http://www.xs4all.nl/~thebeez/4tH

Hans Bezemer

sp...@controlq.com

unread,
Jun 29, 2008, 4:54:40 PM6/29/08
to
On Sat, 28 Jun 2008, Robert Miller wrote:

> I don't know how relevant this might be to modern implementations, but about
> thirty years ago the answer to that question was IIRC sixteen. We had been
> looking at Forth for a 6809 based test system and our software guru (not me)
> found a paper somewhere that identified the minimum wordset from which the
> whole language could be built. It took him only a week to get a limited
> version of FORTH up and running. MVP Forth was the model on which the
> implementation was expanded. It was only another couple of weeks until the
> whole language was implemented. Frequently used words were later recoded in
> assembly language.
>
> Bob

Any links to that paper? The question is not how many, but which ones?

billy

unread,
Jul 1, 2008, 10:29:44 AM7/1/08
to
foxchip <f...@ultratechnology.com> wrote:
> DavidM <nos...@nowhere.com> wrote:
> > Does anyone know of a Forth implementation that:
> >  - is written almost entirely in forth, save for the absolutely
> >    smallest possible number of primitives

> It is a little like asking what is the minimal instruction set.


> It is an academic question not a practical one.  In reality no

The absolute minimum number of instructions for a stack-based language
is two (http://www.eecs.usma.edu/webs/people/okasaki/pubs.html#jfp03).
Unfortunately, with that small of an instruction set there's no room
for numbers (although you can implement numbers as 'Church numerals').

Sorry, just heard the word "academic". Now _this_ is academic.
Definitely not practical.

I'm just finishing off a little interpreter (written in Python, sorry)
for such a language. Somewhat interestingly, it's more fun to program
in than Unlambda. (I'm afraid that establishes a very, VERY low level
of "fun to program".)

-Wm

Coos Haak

unread,
Jul 1, 2008, 11:20:25 AM7/1/08
to
Op 29 Jun 2008 00:23:55 +1200 schreef DavidM:

Lennart Benschop implemented a virtual 32 bit processor with 32 5 bit
opcodes, see:
http://www.xs4all.nl/~lennartb/forth.html

I've implemented a variant of it with 6 bit opcodes and short 5 bit
opcodes. This is a 32 bit virtual processor that runs on 16 bit DOS.

See also various Forth processors in Philip Koopman's essay:
http://www.ece.cmu.edu/%7Ekoopman/stack_computers/index.html
--
Coos

DavidM

unread,
Jul 1, 2008, 4:04:47 PM7/1/08
to
On Sat, 28 Jun 2008 16:03:59 -0700, foxchip wrote:

> There is a version of eForth with primitives in C. You can combine
> those primitives with any other version of eForth you choose.

The eForth developer's site makes no mention of a C version. Google
searches are coming up blank. Do you know where I can find this C verison?

Cheers
Dave

Albert van der Horst

unread,
Jul 1, 2008, 6:56:15 PM7/1/08
to

I would ask for a practical minimum of primitives.
E.g. you probably need a primitive for subtraction.
You could have addition building onto subtraction,
but in practice you are making your Forth more complicated that way.
I have ported ciforth a few times now . An extra primitive here
and there where -- in assembler -- ADD is replaced by XOR or
whatever doesn't cost effort.

The effort in porting is in how to link and make an executable,
and you demand that C solves that for you. In fact you are going
the gforth way.
gforth however has not great simplicity, but aims at being a
system with a good practical usability, and portability.

I guess that your best bet is starting out with FICL and simplify
it in your direction. If someone had done that, he would have
stepped forward by now. In my experience transforming a Forth
system using a regression test like the Hayes suite and a
source control system is not too difficult.


>
>Thanks in advance
>
>Dave
>

Groetjes Albert

--
--
Albert van der Horst, UTRECHT,THE NETHERLANDS
Economic growth -- like all pyramid schemes -- ultimately falters.
albert@spe&ar&c.xs4all.nl &=n http://home.hccnet.nl/a.w.m.van.der.horst

jch

unread,
Jul 1, 2008, 9:55:16 PM7/1/08
to
_____
Hans,

Is there anyone you know who has compiled 4tH on an OpenBSD (3.9 and up)
system? I just pulled down the generic Unix code and will try to do
"make" in the source directory.
--
Regards / JCH

jch

unread,
Jul 1, 2008, 10:06:48 PM7/1/08
to
_____
Yes, the generic Unix Makefile's contents are suitable for an OBSD
system. Compiles with only a few warnings. Runs the 4th executable
just fine. I just don't know yet how to work with it. Need to read the
manual first.
--
Regards / JCH

pml5...@gmail.com

unread,
Jul 1, 2008, 11:22:59 PM7/1/08
to
DavidM wrote:

Well, not quite that, but a few years ago I came across an eforth
implemented in an early version of Perl, 4.0 I think, at the Perl
archive CPAN. That should be straightforward to rework into C. But I
couldn't find it last time I looked for it (also some time back).

Related: I once heard of a highly cut down version of Forth Inc.'s
polyforth called picoforth, with things like DROP and DUP implemented
using @ and ! and temporary variables. Maybe someone can track that
down. P.M.Lawrence.

Stephen Pelc

unread,
Jul 2, 2008, 6:52:40 AM7/2/08
to
On Sat, 28 Jun 2008 18:09:58 -0400, sp...@controlq.com wrote:

>On Sat, 29 Jun 2008, DavidM wrote:
>
>> Hi,
>>
>> Does anyone know of a Forth implementation that:
>>
>> - is written almost entirely in forth, save for the absolutely
>> smallest possible number of primitives
>> - has its handful of primitives written in highly portable C:
>> - can compile for 32-bit or 64-bit, little or big-endian
>> - is ANS compliant
>> - has a good C API, easy to add a small number of extra C
>> primitives
>>
>> I've been working with FICL - it has around 12k lines of C - I'd be
>> curious to work with a Forth with only a few hundred (or even a few
>> dozen) lines of C.
>>
>> Thanks in advance
>>
>> Dave
>
>Having implemented a couple of Forth's in C, it begs the question, just
>what is the minimal wordset with which you can bootstrap a Forth
>implementation? Is there any literature on this?

Most versions of VFX Forth (Windows, Linux, DOS) use about 30 coded
definitions. The DOS versions use more because of hard-coded interrupt
calls. This matches quite well with comments from others.

However, David's first requirement requires a Forth compiler of
some form. If there are only 30 well-defined primitives, I suggest
that their portability is of dubious virtue. Most of the portable
Forths I've seen have become unmaintainable because of the vast
number of conditional compilation switches in the C layer.

The intermediate Forth compiler to glue the Forth source to
the primitives is essentially a Forth cross compiler, however
implemented. That implementation has a big impact on many
qualities of the resulting Forth.

If one can isolate the primitives, and document them, then
producing a primitives files for each implementation should be
a minor exercise. The specification is the major exercise.

Stephen

--
Stephen Pelc, steph...@mpeforth.com
MicroProcessor Engineering Ltd - More Real, Less Time
133 Hill Lane, Southampton SO15 5AF, England
tel: +44 (0)23 8063 1441, fax: +44 (0)23 8033 9691
web: http://www.mpeforth.com - free VFX Forth downloads

The Beez'

unread,
Jul 2, 2008, 7:54:53 AM7/2/08
to
On 2 jul, 03:55, jch <j...@nowhere.net> wrote:
> Is there anyone you know who has compiled 4tH on an OpenBSD (3.9 and up)
> system?  I just pulled down the generic Unix code and will try to do
> "make" in the source directory.
I don't get an update on each and every release, but yes, 4tH is known
to work on BSD (amoungst others). Most *nix systems do just fine.

Hans Bezemer

pml5...@gmail.com

unread,
Jul 2, 2008, 9:20:30 AM7/2/08
to
pml540...@gmail.com wrote:

> DavidM wrote:
>
> > On Sat, 28 Jun 2008 16:03:59 -0700, foxchip wrote:
> >
> > > There is a version of eForth with primitives in C. You can combine
> > > those primitives with any other version of eForth you choose.
> >
> > The eForth developer's site makes no mention of a C version. Google
> > searches are coming up blank. Do you know where I can find this C verison?
>
> Well, not quite that, but a few years ago I came across an eforth
> implemented in an early version of Perl, 4.0 I think, at the Perl
> archive CPAN. That should be straightforward to rework into C. But I
> couldn't find it last time I looked for it (also some time back).

Aha! This may be it or something like it, though the one I came across
had more comments and acknowledged that it was based on eforth:
http://search.cpan.org/~petergal/PGForth1.3/Language/PGForth.pm -
P.M.Lawrence.

foxchip

unread,
Jul 2, 2008, 10:20:12 AM7/2/08
to
On Jul 1, 1:04 pm, DavidM <nos...@nowhere.com> wrote:
> The eForth developer's site makes no mention of a C version. Google
> searches are coming up blank. Do you know where I can find this C verison?

There is a minimum for what has to be written in Forth, and a minimum
for
has to be written in something other than Forth. Those minimum are
zero.
Bill prefers to use a Forth metacompiler and Forth assembler for
primitives.

Goggle "c eforth" for primitives in assembler or binary for linking in
a
C build, threaded Forth converted to C, and C library hooks for
further
expansion with more C functions.

Best Wishes

astrobe

unread,
Jul 2, 2008, 12:42:41 PM7/2/08
to

It's kindof a Forth dialect but 4IM (http://astrobe.byethost31.com) is
<1KLOC for its core VM (written in C). It has about 50 C primitives.
The whole C source is about 5KLOC. I think it will slim down in the
future. It has (IMHO :) a good C API; I've already written a wrapper
for a GUI lib, sockets, serial comm both for Linux ans Windows, plus
other ones for proprietary libraries. And I'm currently writting a
wrapper for the Irrlicht 3D engine.

Amicalement,
Astrobe

Brad Eckert

unread,
Jul 2, 2008, 1:08:29 PM7/2/08
to
On Jun 28, 5:23 am, DavidM <nos...@nowhere.com> wrote:
>
>  - is written almost entirely in forth, save for the absolutely
>    smallest possible number of primitives
>  - has its handful of primitives written in highly portable C:
>     - can compile for 32-bit or 64-bit, little or big-endian
>  - is ANS compliant
>  - has a good C API, easy to add a small number of extra C
>    primitives
>
There's Fminus on Sourceforge. It's a Forth to C (or asm) cross
compiler that generates a VM that executes bytecodes. CODE words
consist of C text. The resulting VM is one file containing a bytecode
ROM and a switch statement of CODE primitives. If you're trying to put
a minimal-footprint Forth on an embedded target, it may be useful.

Brad

Rod Pemberton

unread,
Jul 2, 2008, 5:10:25 PM7/2/08
to
<sp...@controlq.com> wrote in message
news:alpine.BSF.1.10.0...@saddam.controlq.com...
> On Sat, 29 Jun 2008, DavidM wrote:
> > Does anyone know of a Forth implementation that:
> >
> > - is written almost entirely in forth, save for the absolutely
> > smallest possible number of primitives
> > - has its handful of primitives written in highly portable C:
> > - can compile for 32-bit or 64-bit, little or big-endian
> > - is ANS compliant
> > - has a good C API, easy to add a small number of extra C
> > primitives
> >
> > I've been working with FICL - it has around 12k lines of C - I'd be
> > curious to work with a Forth with only a few hundred (or even a few
> > dozen) lines of C.
> >
>
> Having implemented a couple of Forth's in C, it begs the question, just
> what is the minimal wordset with which you can bootstrap a Forth
> implementation?

Sorry for the delay, gone for a few... Anyway, I've been collecting info on
that. I'll post the relevant info below my sig. The counts should be taken
as "being 95-99% close" not as "100% exact". (IIRC, I posted some of this
previously) First, the FORTH primitives info I have is listed. (I'm
missing info on C. Moore's machine FORTH and I haven't counted the complete
ANS wordset...). Second, there are some other languages, API's, etc., to
provide a rough comparison. Third, I've listed the primitives used by a few
smaller FORTHs, LISPs, BF, etc.

> Is there any literature on this?

No, not that I'm aware of. The info I've provided below is the most
complete I've ever seen. I became interested in the smallest set of
functionality needed to implement C, but I found much more on FORTH
"primitives" at first.

HTH,

Rod Pemberton
PS. I have not kept track of the FORTH's below. But, if you need me to try
to relocate it, just say so.

---
FORTH Primitives Comparison (use a fixed width font)
---
3 primitives - Frank Sargent's "3 Instruction Forth"
9 primitives - Mark Hayes theoretical minimal Forth bootstrap
9,11 primitives - Mikael Patel's Minimal Forth Machine (9 minimum, 11 full)
13 primitives - theoretical minimum for a complete FORTH (Brad Rodriguez)
16,29 primitives - C. Moore's word set for the F21 CPU (16 minimum, 29 full)
20 primitives - Philip Koopman's "dynamic instruction frequencies"
23 primitives - Mark Hayes MRForth
25 primitives - C. Moore's instruction set for MuP21 CPU
36 primitives - Dr. C.H. Ting's eForth, a highly portable forth
46 primitives - GNU's GFORTH for 8086
58-255 functions - FORTH-83 Standard (255 defined, 132 required, 58 nucleus)
60-63 primitives - considered the essence of FORTH by C. Moore (unknown)
72 primitives - Brad Rodriguez's 6809 CamelForth
74-236 functions - FORTH-79 Standard (236 defined, 147 required, 74 nucleus)
94-229 functions - fig-FORTH Std. (229 defined, 117 required, 94 level zero)
133-? functions - ANS-FORTH Standard (? defined, 133 required, 133 core)
200 functions - FORTH 1970, the original Forth by C. Moore
240 functions - MVP-FORTH (FORTH-79)
~1000 functions - F83 FORTH
~2500 functions - F-PC FORTH

FIXME 27 ? - C. Moore's MachineForth

For comparison:
---
8 commands - BrainFuck (small,Turing complete language)
8 primitives - Stutter LISP
8 primitives - LISP generic
11 functions - OS functions Ritchie & Thompson PDP-7 and/or PDP-11 Unix
14 primitives - LISP McCarthy based
18 functions - OS functions required by P.J. Plauger's Standard C
Library
19 functions - OS functions required by Redhat's newlib C library
28 opcodes - LLVA - Low Level Virtual instruction set Architecture
51-56 functions - CP/M 1.3 (36-41 BDOS, 15 BIOS)
56 functions - CP/M 2.2 (39 BDOS, 17 BIOS)
40 syscalls - Linux v0.01 (67 total, 13 unused, 14 minimal, 40
complete)
71 opcodes - LLVM - Low Level Virtual Machine instructions
92+ functions - MP/M 2.1 (92 BDOS, ? BIOS)
102 functions - CP/M 3.0 (69 BDOS, 33 BIOS)
~120 functions - OpenWATCOM v1.3, calls - DOS, BIOS, DPMI for PM DOS apps.
150 syscalls - GNU HURD kernel
170 functions - DJGPP v2.03, calls - DOS, BIOS, DPMI for PM DOS apps.
206 bytecodes - Java Virtual Machine bytecodes
290 syscalls - Linux Kernel 2.6.17 (POSIX.1)

eForth primitives (9 optional)
----
doLIT doLIST BYE EXECUTE EXIT next ?branch branch ! @ C! C@ RP@ RP! R> R@ >R
SP@ SP! DROP DUP SWAP OVER 0< AND OR XOR UM+ TX!
?RX !IO $CODE $COLON $USER D$ $NEXT COLD IO?

9 MRForth bootstrap theoretical
----
@ ! + AND XOR (URSHIFT) (LITERAL) (ABORT) EXECUTE

9 Minimal Forth (3 optional)
----
>r r> 1+ 0= nand @ dup! execute exit
drop dup swap

23 MRForth primitives
----
C@ C! @ ! DROP DUP SWAP OVER $>$R R$>$ + AND OR XOR (URSHIFT) 0$<$ 0=
(LITERAL) EXIT (ABORT) (EMIT) (KEY)

20 Koopman high execution, Dynamic Freq.
----
CALL EXIT EXECUTE VARIABLE USER LIT CONSTANT 0BRANCH BRANCH I @ C@ R> >R
SWAP DUP ROT + = AND

46 Gforth
----
:DOCOL :DOCON :DODEFER :DOVAR :DODOES ;S BYE EXECUTE BRANCH ?BRANCH LIT @ !
C@ C! SP@ SP! R> R@ >R RP@ RP! + - OR XOR AND 2/ (EMIT) EMIT? (KEY) (KEY?)
DUP 2DUP DROP 2DROP SWAP OVER ROT -ROT UM* UM/MOD LSHIFT RSHIFT 0= =

36 eForth
-------
BYE ?RX TX! !IO doLIT doLIST EXIT EXECUTE next ?branch branch ! @ C! C@ RP@
RP! R> R@ >R SP@ SP! DROP DUP SWAP OVER 0< AND OR XOR UM+ $NEXT D$ $USER
$COLON $CODE

BrainFuck
-------
> < + - . , [ ]

Stutter LISP
----
car cdr cons if set equal lambda quote

generic LISP
----
atom car cdr cond cons eq lambda quote

LISP, McCarthy based
----
and atom car cdr cond cons eq eval lambda nil quote or set t

DavidM

unread,
Jul 2, 2008, 6:07:26 PM7/2/08
to
Wow, what a goldmine - thanks heaps for that! :)

Dave

Elizabeth D Rather

unread,
Jul 2, 2008, 6:31:33 PM7/2/08
to
Rod Pemberton wrote:
...

> Sorry for the delay, gone for a few... Anyway, I've been collecting info on
> that. I'll post the relevant info below my sig. The counts should be taken
> as "being 95-99% close" not as "100% exact". (IIRC, I posted some of this
> previously) First, the FORTH primitives info I have is listed. (I'm
> missing info on C. Moore's machine FORTH and I haven't counted the complete
> ANS wordset...). Second, there are some other languages, API's, etc., to
> provide a rough comparison. Third, I've listed the primitives used by a few
> smaller FORTHs, LISPs, BF, etc.
...

> ---
> FORTH Primitives Comparison (use a fixed width font)
> ---
> 3 primitives - Frank Sargent's "3 Instruction Forth"
> 9 primitives - Mark Hayes theoretical minimal Forth bootstrap
> 9,11 primitives - Mikael Patel's Minimal Forth Machine (9 minimum, 11 full)
> 13 primitives - theoretical minimum for a complete FORTH (Brad Rodriguez)
> 16,29 primitives - C. Moore's word set for the F21 CPU (16 minimum, 29 full)
> 20 primitives - Philip Koopman's "dynamic instruction frequencies"
> 23 primitives - Mark Hayes MRForth
> 25 primitives - C. Moore's instruction set for MuP21 CPU
> 36 primitives - Dr. C.H. Ting's eForth, a highly portable forth
> 46 primitives - GNU's GFORTH for 8086
> 58-255 functions - FORTH-83 Standard (255 defined, 132 required, 58 nucleus)
> 60-63 primitives - considered the essence of FORTH by C. Moore (unknown)
> 72 primitives - Brad Rodriguez's 6809 CamelForth
> 74-236 functions - FORTH-79 Standard (236 defined, 147 required, 74 nucleus)
> 94-229 functions - fig-FORTH Std. (229 defined, 117 required, 94 level zero)
> 133-? functions - ANS-FORTH Standard (? defined, 133 required, 133 core)
> 200 functions - FORTH 1970, the original Forth by C. Moore
> 240 functions - MVP-FORTH (FORTH-79)
> ~1000 functions - F83 FORTH
> ~2500 functions - F-PC FORTH
>
> FIXME 27 ? - C. Moore's MachineForth

Bear in mind that the fact that a word is designated as "nucleus" or
"core" in the various standards has nothing to do with whether it's a
"primitive" in the sense of a word that's defined in assembler, C, or
another language and necessary to implement enough of the Forth VM to
function.

Like MPE, FORTH, Inc. writes its systems entirely in Forth and Forth
assembler, using a cross-compiler to bootstrap a new platform. There
are relatively few assembler "primitives" in our system, but we base the
decision as to which should be coded in assembler only partly on
necessity, and mostly on performance issues. And even that is
influenced by the fact that we're compiling to machine code with an
optimizer.

We confine these words (plus other platform-specific issues such as
configuration, startup, etc.) to specific files, so the chore of porting
to a new platform just requires recoding those files and the rest of the
files that make up the system are unchanged.

Cheers,
Elizabeth

--
==================================================
Elizabeth D. Rather (US & Canada) 800-55-FORTH
FORTH Inc. +1 310.999.6784
5959 West Century Blvd. Suite 700
Los Angeles, CA 90045
http://www.forth.com

"Forth-based products and Services for real-time
applications since 1973."
==================================================

Rod Pemberton

unread,
Jul 2, 2008, 8:27:38 PM7/2/08
to

"Elizabeth D Rather" <era...@forth.com> wrote in message
news:a5ednTzymc1bnPHV...@supernews.com...

Yes, thanks for the reminder. But, I believe the ones with "core" and
"nucleus" do say "functions" instead of "primitives"... While that's
probably not FORTH proper terminology, I didn't proclaim them to be
"primitives".


Rod Pemberton

John Passaniti

unread,
Jul 3, 2008, 7:19:57 PM7/3/08
to
Rod Pemberton wrote:
> 3 primitives - Frank Sargent's "3 Instruction Forth"

That is a Forth only in the most abstract sense possible.

But I do have one to add to your list that I've mentioned before in the
past. The 1992 Obfuscated C Code Contest offered "buzzard.2" which
bootstraps a small near-Forth language from 16 primitives (13 visible, 3
internal) implemented in about 700 characters of C code. This
implements a virtual machine which then bootstraps into the near-Forth
from source that, except for the first few lines, is Forth. It's
interesting to see how some of the lowest-level primitives are built-up.

Removing the obfuscation also removes the joy and fun of the contest,
but leaves a small virtual machine that can be loaded on embedded
systems very easily and provide a good Forth-like environment for
interactively bringing hardware up. I wouldn't use it for anything
serious because like most of these reduced-primitive Forths, it's pretty
slow. But it gets the job done.

You can find the code here:

http://www.ioccc.org/years.html#1992

foxchip

unread,
Jul 4, 2008, 3:11:53 PM7/4/08
to
On Jul 2, 2:10 pm, "Rod Pemberton" <do_not_h...@nohavenot.cmm> wrote:
> (I'm missing info on C. Moore's machine FORTH

A complete history with details of the evolution of machine Forth
dialects through Sh-Boom, P20, P21, P8, P32, I21, f21, and c18 can
be found at or linked from
http://www.ultratechnology.com
latest on c18 at http://www.IntellaSys.net

Sh-Boom had eight-bit opcodes and P20 on used five-bit opcodes. P20
had skip opcodes which were placed by 10-bit paged branches in P21
which had 25 opcodes. F21 changed 2/ to change bit-19 and added
more branching modes, addressing opcodes and coprocessors expanding
the instruction set to 28 including !p+. C18 added the B register
opcodes, added 2/ to the +* opcode, and added ;: coroutine.

Different opcode names were used in different software. Chuck's
original P21 machineForth used more one character names than are
in common use today. I always found writing down and grouping
the opcodes to be a useful exercise.
~ means not present in an instruction set.

orig p21 F21 C18 c18-colorforth

\ stack
. nop nop . .
t dup dup dup dup
t! drop drop drop drop
s over over over over

\ math-logic
nd and and and and
or or xor xor or \ all exclusive
- com com not -
2/ 2/ 2/ 2/ 2/ \ top two bits don't change on P21
2* 2* 2* 2* 2*
+ + + + +
* +* +* +* +* \ C18 includes 2/

\ addressing
a a a@ a@ a
a! a! a! a! a!
! ! !a !a !
!+ !+ !a+ !a+ !+
@ @ @a @a @
@+ @+ @a+ @a+ @+

n lit # @p+ @p+
~ ~ !p+ !p+ !p+

~ ~ @r+ ~ ~
~ ~ !r+ ~ ~

~ ~ ~ b! b!
~ ~ ~ @b @b
~ ~ ~ !b !b

\ return-stack
r @r r> pop pop
!r !r >r push push

\ branch
T0 T0 if if if \ until while
C0 CO -if -if -if \ -until -while
: call call call call
jump jump jump jump jump \ else
; ; ; ; ;
~ ~ ~ ;: ;:
~ ~ ~ next next
~ ~ ~ unext unext

I still sometimes use the original machineForth dialect
for brevity, or a version where most Forth words are
one character.

Older Forth hardware design is covered in Koopman's online
book on Stack Machines.

Best Wishes

jacko

unread,
Jul 9, 2008, 4:06:04 PM7/9/08
to
hi

Very comprehensive. MID4th is upto 70 primitives, but if files is
removed then this reduces somewhat. I do intend to reduce this number
of the comming weeks, as the aim is to make it portable to custom
hardware as easily as possible. I will place boot options to control
which sections of the dictionary are compiled in.

I think @1+ 1-! AND + XOR as register to register one and two argument
functions to be defined in C is about the simplest virtual machine.
Even though - is kind of more primitive than +, + can be used to do
LSHIFT. a+a=a<<1 but a-a=0

U/MOD and U* can be made faster as an example by direct coding, but
both can be built from U*/ (as I call it) which can be found described
on http://indi.hpsdr.com/system32.html and I am sure there are other
refactorings possible.

I hope to get MID4th down to 64 primitives with a full feature set.
It uses the fact that 2 buffers exsist in high memory to indicate
primitives, but this does prevent execution of xts from these buffers.
It does make 16 bit truncation of the dictionary for storage much
easier though when a 64K address space is used.


cheers

0 new messages