I stumbled upon an interesting article:
http://www.rcollins.org/articles/pmbasics/tspec_a1_doc.html

I wonder whether were there any Forths, that were
switching from DOS' real mode to operate in protected mode
(and ev. their BYE meant getting back to DOS in its real mode)?

Zbig schrieb am Donnerstag, 5. Oktober 2023 um 22:18:10 UTC+2:
> I stumbled upon an interesting article:
> http://www.rcollins.org/articles/pmbasics/tspec_a1_doc.html
>
> I wonder whether were there any Forths, that were
> switching from DOS' real mode to operate in protected mode
> (and ev. their BYE meant getting back to DOS in its real mode)?

Back then Docker containers had not yet been invented.. ;-)

> > I stumbled upon an interesting article:
> > http://www.rcollins.org/articles/pmbasics/tspec_a1_doc.html
> >
> > I wonder whether were there any Forths, that were
> > switching from DOS' real mode to operate in protected mode
> > (and ev. their BYE meant getting back to DOS in its real mode)?
> Back then Docker containers had not yet been invented.. ;-)

How come? There were „native Forths” (contrary to „hosted” ones).

On Thursday, October 5, 2023 at 10:18:10 PM UTC+2, Zbig wrote:
> I stumbled upon an interesting article:
> http://www.rcollins.org/articles/pmbasics/tspec_a1_doc.html
>
> I wonder whether were there any Forths, that were
> switching from DOS' real mode to operate in protected mode
> (and ev. their BYE meant getting back to DOS in its real mode)?

Just before the serious ansification efforts started, there was F-4TH
and FFORTH. These where wildly excentric Forths with 3 byte addresses
an 2 byte data. They used a "DOS-extender" to put the system in
protected mode, and then segmented addresses could be used to access
4MB or more of memory. When the Forth standard allowed non-16 bit
implementations I switched to full 32bits.

I had a look just now but can't find the iForth docs from the
interbellum.

-marcel

On Thursday, October 5, 2023 at 11:53:17 PM UTC+2, Marcel Hendrix wrote:
> On Thursday, October 5, 2023 at 10:18:10 PM UTC+2, Zbig wrote:
> > I stumbled upon an interesting article:
> > http://www.rcollins.org/articles/pmbasics/tspec_a1_doc.html
> >
> > I wonder whether were there any Forths, that were
> > switching from DOS' real mode to operate in protected mode
> > (and ev. their BYE meant getting back to DOS in its real mode)?
> Just before the serious ansification efforts started, there was F-4TH
> and FFORTH. These where wildly excentric Forths with 3 byte addresses
> an 2 byte data. They used a "DOS-extender" to put the system in
> protected mode, and then segmented addresses could be used to access
> 4MB or more of memory. When the Forth standard allowed non-16 bit
> implementations I switched to full 32bits.
>
> I had a look just now but can't find the iForth docs from the
> interbellum.
>
> -marcel

Same here. I still release a 4tH version for DOS with a CWSDPMI DOS extender.
AFAIK Gforth 0.30 used the very same method - compiled by DJGPP.

4tH however, never used 16bits. I hated Forth-83. Forth almost lost me because
of that abomination.. The first version of 4tH was almost entirely Forth-79.. That
was such a fun, happy standard! Forth-83 was schizophrenic, psychotic
and paranoid.

Fortunately, they locked it up, threw away the key and forgot all about it.

And ANS? ANS is just depressing. But with a few Xanax it can make it through the
day. If its OCD isn't too bad, that is.

Hans Bezemer

In article <c8546325-c67b-4381-b3ea-16fb02e871f3n@googlegroups.com>,
Zbig <zbigniew2011@gmail.com> wrote:
>I stumbled upon an interesting article:
>http://www.rcollins.org/articles/pmbasics/tspec_a1_doc.html
>
>I wonder whether were there any Forths, that were
>switching from DOS' real mode to operate in protected mode
>(and ev. their BYE meant getting back to DOS in its real mode)?

I did this in 2002. It only makes sense if MSDOS system abound.
One of the Forths generated by the ciforth-factory is a
32 bit Forths doing exactly what you propose, and one time
I had a system that booted into such a 32 but Forth and
had a million blocks available from a hd.

2023:
I have bought multi-core 64 bits risc-V and 64 bit ARM
system with a fully 1000+ data for the SOC
( H6, RK3399 etc.).
I can't be excited about MSDOS anymore.

Groetjes Albert
--
Don't praise the day before the evening. One swallow doesn't make spring.
You must not say "hey" before you have crossed the bridge. Don't sell the
hide of the bear until you shot it. Better one bird in the hand than ten in
the air. First gain is a cat spinning. - the Wise from Antrim -

Thanks, guys.
Yes, DOS presently is one more „retro-system”, still
it keeps its merits as learning platform.
And the solution presented by Mr. Collins was
something new to me.

On Friday, October 6, 2023 at 7:28:01 AM UTC-7, Zbig wrote:
> Thanks, guys.
> Yes, DOS presently is one more „retro-system”, still
> it keeps its merits as learning platform.

In the early 90s, 32-bit UR/Forth from LMI running under a
DOS-extender was the flagship of the Forth community.
UR/Forth was what really kept Forth competitive with C.
I did some benchmarking of 16-bit UR/Forth against Borland
Turbo-C (Small memory-model) and found the speed to be
the same. These benchmarks were my own programs that
did recursive-descent searches and took several minutes to run.
I never benchmarked 32-bit UR/Forth against C because I
didn't have access to 32-bit UR/Forth prior to working at Testra.
At Testra I wrote MFX in 32-bit UR/Forth.
There was no other Forth available at the time that would
have been capable of supporting MFX. UR/Forth was great!
Even today, SwiftForth and VFX would not be capable of
supporting MFX without significant upgrade to fix the problems
of ANS-Forth --- for example, the wordlists in ANS-Forth are
screwed up (I fix this in the novice-package to provide them
with the same features as Forth-83 vocabularies) --- as another
example, FIND and tick are broken in ANS-Forth (I fix this in
the novice-package to make them work again like in Forth-83).

The purpose of ANS-Forth was to declare UR/Forth non-standard.
ANS-Forth killed UR/Forth and put LMI out of business, and this
killed Forth --- after UR/Forth was gone the Forth community
no longer had a Forth system that was competitive with C
and capable of supporting commercial applications.

> And the solution presented by Mr. Collins was
> something new to me.

Are you referring to the troll Rick Collins? WTF???
He is not in this thread.
He has never presented any kind of solution to anything
due to the fact that he is not a programmer.

> > Thanks, guys.
> > Yes, DOS presently is one more „retro-system”, still
> > it keeps its merits as learning platform.
> In the early 90s, 32-bit UR/Forth from LMI running under a
> DOS-extender was the flagship of the Forth community.
> UR/Forth was what really kept Forth competitive with C.

No trace of UR/Forth anywhere in the Internet (I once was
seaching for maybe some "shareware" version for download).
Nothing.

I recall trying to contact its author some 3 years ago, but even
to find his e-mail address turned out to be not possible.

> > And the solution presented by Mr. Collins was
> > something new to me.
> Are you referring to the troll Rick Collins? WTF???

I meant the author of the linked article — Robert Collins.

On Monday, October 9, 2023 at 5:31:06 PM UTC+2, Zbig wrote:
[..]
> No trace of UR/Forth anywhere in the Internet (I once was
> seaching for maybe some "shareware" version for download).
> Nothing.
>
> I recall trying to contact its author some 3 years ago, but even
> to find his e-mail address turned out to be not possible.

He's on LinkedIn and his profile leaves no doubt:

"Contributing Editor
Dr. Dobb's Journal of Software Tools
1982 - 2002 · 20 yrs
Author of numerous feature articles as well as the regular
columns "16-bit Software Toolbox" and "Programmer's Bookshelf.""

-marcel

> He's on LinkedIn and his profile leaves no doubt:

Thank you, I'll take a look.

On Monday, October 9, 2023 at 8:31:06 AM UTC-7, Zbig wrote:
> > In the early 90s, 32-bit UR/Forth from LMI running under a
> > DOS-extender was the flagship of the Forth community.
> > UR/Forth was what really kept Forth competitive with C.
> No trace of UR/Forth anywhere in the Internet (I once was
> seaching for maybe some "shareware" version for download).
> Nothing.

UR/Forth was a commercial product. I paid $400 for the 16-bit version
in the early 90s. I never bought the 32-bit version. I remember that I
wrote an improved software floating-point package (the package
that came with UR/Forth had bugs). I tried to sell this to Ray Duncan
but he wouldn't pay cash --- he did offer to give me a copy of the
32-bit UR/Forth but I declined --- later on I got a job at Testra and
used their copy of the 32-bit UR/Forth.

When LMI was killed off by ANS-Forth, Testra obtained the
source-code for UR/Forth from Ray Duncan, but they had to sign
an NDA. Later on Testra converted UR/Forth from running under
a 32-bit DOS-extender to running under 32-bit Windows. They did
this so they could continue running my MFX and other legacy
UR/Forth software --- they continue using UR/Forth even now in 2023.
The NDA prevents Testra from releasing MFX open-source because
MFX only runs on UR/Forth. The NDA also prevents Testra from
passing their Windows version of UR/Forth on to the handful
of other people who obtained UR/Forth source-code.

Elizabeth Rather killed UR/Forth by declaring it non-standard
by fiat. Her expectation was that the large UR/Forth community
would switch to PolyForth, or patiently wait years for SwiftForth
to come out (and more years for the bugs in SwiftForth to get
fixed so that SwiftForth was usable). This didn't happen.
The large UR/Forth community switched to C programming.

ANS-Forth was a purely destructive contribution to Forth.
If Elizabeth Rather hadn't been so evil, striving to destroy
to competition, UR/Forth would have continued as the
flagship of the Forth community and would have been
upgraded to run under Windows so Forth could continue
to be competitive with C under the new OS.
Elizabeth Rather is personally to blame for the death of Forth.

On 10 Oct 23 at group /comp/lang/forth in article d481c139-c737-485d-a659-854609bda3afn@googlegroups.com
<hughaguilar96@gmail.com> (Hugh Aguilar) wrote:

> On Monday, October 9, 2023 at 8:31:06?AM UTC-7, Zbig wrote:
>>> In the early 90s, 32-bit UR/Forth from LMI running under a
>>> DOS-extender was the flagship of the Forth community.
>>> UR/Forth was what really kept Forth competitive with C.
>> No trace of UR/Forth anywhere in the Internet (I once was
>> seaching for maybe some "shareware" version for download).
>> Nothing.

> UR/Forth was a commercial product. I paid $400 for the 16-bit version
> in the early 90s. I never bought the 32-bit version. I remember that I
> wrote an improved software floating-point package (the package
> that came with UR/Forth had bugs). I tried to sell this to Ray Duncan
> but he wouldn't pay cash --- he did offer to give me a copy of the
> 32-bit UR/Forth but I declined --- later on I got a job at Testra and
> used their copy of the 32-bit UR/Forth.

I met Ray Duncan 2 or 3 times personally in the 80s in Germany, even in my
office.

I purchased different LMI products, for a lot of money, especially a lot
of MetaCompilers. I made more of hundred projects for different industrial
target. Z80, 64180, RTX-2000, 8031, 80535, 80c166, 80c192, 68000, 68HC16,
8086, 6802...

Not only the the UR/Forth FP had bugs, there where a lot of more bugs in
MC. Especially buggy was the 80C166 and 80535.

I fixed more than a dozend bugs which I was able to fix/circumvent.
And demanded Ray to fix the bugs in the MC Kernal. But Ray denied to fix
any and told the German Distributor, that he should piss me and refund my
money for the 80C166. Ray didn't wan't to discuss any longer with me, he
had no clue, what an engineer needs nor how to make really professionell
SW for industrie. Ray was the 'GURU' and Gurus never make any mistakes!

I just wan't quality for my quality money, no more, no less!
Not beeing pissed off!

Ok, Ray has a PhD as children doctor, no engineer, no electronics.
Nor he had any idea of industrial engineering.

He made a mistake while selling me some MC, one disk contains the complete
source for the MC Kernal. Poor documentation and quite a hack. I fiddled
several days to repair the bugs and recompiled 'my' MC-kernals.
So I could survive with my engineering office, delivering industrial
controllers and projects. :)

> When LMI was killed off by ANS-Forth, Testra obtained the
> source-code for UR/Forth from Ray Duncan, but they had to sign
> an NDA.

I had one customer, who also bought the source code for PC-Forth, RTX-2000
MC... Just in case, that LMI went of the market. Good decission :)

> Elizabeth Rather killed UR/Forth by declaring it non-standard
> by fiat. Her expectation was that the large UR/Forth community
> would switch to PolyForth, or patiently wait years for SwiftForth
> to come out (and more years for the bugs in SwiftForth to get
> fixed so that SwiftForth was usable). This didn't happen.
> The large UR/Forth community switched to C programming.

And LMI died due to the arrogant behaivior of Ray 'Guru' Duncan.

Ray forged his own coffin nails. Maybe Beth handled him a hammer :]

Saludos (an alle Vernünftigen, Rest sh. sig)
Wolfgang

--
Ich bin in Paraguay lebender Trollallergiker :) reply Adresse gesetzt!
Ich diskutiere zukünftig weniger mit Idioten, denn sie ziehen mich auf
ihr Niveau herunter und schlagen mich dort mit ihrer Erfahrung! :p
(lt. alter usenet Weisheit) iPod, iPhone, iPad, iTunes, iRak, iDiot

On 11/10/2023 9:44 am, Wolfgang Allinger wrote:
>
> On 10 Oct 23 at group /comp/lang/forth in article d481c139-c737-485d-a659-854609bda3afn@googlegroups.com
> <hughaguilar96@gmail.com> (Hugh Aguilar) wrote:
> ...
>> UR/Forth was a commercial product. I paid $400 for the 16-bit version
>> in the early 90s. I never bought the 32-bit version. I remember that I
>> wrote an improved software floating-point package (the package
>> that came with UR/Forth had bugs). I tried to sell this to Ray Duncan
>> but he wouldn't pay cash --- he did offer to give me a copy of the
>> 32-bit UR/Forth but I declined --- later on I got a job at Testra and
>> used their copy of the 32-bit UR/Forth.
> ...
> Not only the the UR/Forth FP had bugs, there where a lot of more bugs in
> MC. Especially buggy was the 80C166 and 80535.

Out of curiosity does anyone recall what were the LMI F/P bugs? Can't
say I used it though I did disassemble it - both software and FPU versions.
I noticed formatted output truncated rather than rounded but perhaps not
surprising given the vintage. AFAICT FP never featured highly in Forth.

On Tuesday, October 10, 2023 at 3:50:20 PM UTC-7, Wolfgang Allinger wrote:
> I purchased different LMI products, for a lot of money, especially a lot
> of MetaCompilers. I made more of hundred projects for different industrial
> target. Z80, 64180, RTX-2000, 8031, 80535, 80c166, 80c192, 68000, 68HC16,
> 8086, 6802...
>
> Not only the the UR/Forth FP had bugs, there where a lot of more bugs in
> MC. Especially buggy was the 80C166 and 80535.
>
> I fixed more than a dozend bugs which I was able to fix/circumvent.
> And demanded Ray to fix the bugs in the MC Kernal. But Ray denied to fix
> any and told the German Distributor, that he should piss me and refund my
> money for the 80C166. Ray didn't wan't to discuss any longer with me, he
> had no clue, what an engineer needs nor how to make really professionell
> SW for industrie. Ray was the 'GURU' and Gurus never make any mistakes!

I found a bug in UR/Forth. It has a low-level undocumented word called
HEADER that generated a header and was used by colon etc.. It over-wrote
memory beyond the memory that it was supposed to use. I told Ray Duncan
about this but he refused to fix it. He said that fixing it would slow down
compile speed and that it didn't matter because when colon etc. use it,
the memory beyond is unused (when I used it though, that memory was
used and over-writing it could cause a hard-to-find bug). I worked around
the problem by using a temporary string on the circular-buffer. This was
slower, but it was reliable. This was for my 65c02 cross-compiler. Later
at Testra at the job interview I was asked if I was familiar with UR/Forth
and I told them about this bug as an example of how familiar I was.
John Hart already knew about the bug and had also told Ray Duncan
about it to no avail. So I agree that Ray Duncan was less than professional..
Still though --- UR/Forth was by far the most professional quality Forth
available and was carrying the Forth community --- PolyForth was of such
low quality that it compared badly against public-domain and $20 shareware
Forth systems (such as those from Tom Zimmer).

Testra's motion-control board was based on the Dallas 80c320 and they
programmed it using a cross-compiler from LMI. I never was involved
in that --- when I was hired they started the MiniForth project that I wrote
MFX for --- I've never used any cross-compiler other than my own, but from
what I've read about them I consider them all to be inferior to what I wrote.
I always write code from scratch without looking at what already exists.
I always feel confident that what I write is superior to anything that already exists.

> I fixed more than a dozend bugs which I was able to fix/circumvent.
> And demanded Ray to fix the bugs in the MC Kernal. But Ray denied to fix
> any and told the German Distributor, that he should piss me and refund

Hmm... so maybe I shouldn't bother Mr. Duncan after all.

Could you list a few of these bugs? I mean as a "memento",
what should be avoided when creating a Forth compiler.

On 11/10/2023 5:31 pm, Zbig wrote:
>> I fixed more than a dozend bugs which I was able to fix/circumvent.
>> And demanded Ray to fix the bugs in the MC Kernal. But Ray denied to fix
>> any and told the German Distributor, that he should piss me and refund
>
> Hmm... so maybe I shouldn't bother Mr. Duncan after all.

Not if you're one of the "FIG-public domain-hacker cabal" that
"systematically destroyed" the Forth market :)

> Not if you're one of the "FIG-public domain-hacker cabal" that
> "systematically destroyed" the Forth market :)

I guess it was a stance of the type „priest of secret lore” and the
desire to maintain (kind of) a monopoly? ;)
400 USD even today it's still not that cheap. Yes, I'm aware it
depends „what for” — but from what I see there were real problems
to be solved/fixed.

"I always write code from scratch without looking at what already exists."

"I always feel confident that what I write is superior to anything that already exists. "

Who is this guy - I must hire him?

In article <GQbfNFvEQoB@allinger-307049.user.uni-berlin>,
Wolfgang Allinger <all2001@gmx.de> wrote:
<SNIP>
>He made a mistake while selling me some MC, one disk contains the complete
>source for the MC Kernal. Poor documentation and quite a hack. I fiddled
>several days to repair the bugs and recompiled 'my' MC-kernals.
>So I could survive with my engineering office, delivering industrial
>controllers and projects. :)

Is their any value to this compiler? Should you preserve it for posterity?
The product is no longer available, so you could publish it in Germany.

>Saludos (an alle Vern�nftigen, Rest sh. sig)
>Wolfgang
>
>--

Groetjes Albert
--
Don't praise the day before the evening. One swallow doesn't make spring.
You must not say "hey" before you have crossed the bridge. Don't sell the
hide of the bear until you shot it. Better one bird in the hand than ten in
the air. First gain is a cat spinning. - the Wise from Antrim -

On 11 Oct 23 at group /comp/lang/forth in article 371de35d-1688-4519-9bb6-dc6465c1a1a7n@googlegroups.com
<zbigniew2011@gmail.com> (Zbig) wrote:

>> I fixed more than a dozend bugs which I was able to fix/circumvent.
>> And demanded Ray to fix the bugs in the MC Kernal. But Ray denied to fix
>> any and told the German Distributor, that he should piss me and refund

> Hmm... so maybe I shouldn't bother Mr. Duncan after all.

> Could you list a few of these bugs? I mean as a "memento",
> what should be avoided when creating a Forth compiler.

No, I don't remember well. It was in the 80s!
But I remember the primadonna Ray very well :(

However, his book about MS-DOS, was excellent, his SW not.

Saludos (an alle Vernünftigen, Rest sh. sig)
Wolfgang

--
Ich bin in Paraguay lebender Trollallergiker :) reply Adresse gesetzt!
Ich diskutiere zukünftig weniger mit Idioten, denn sie ziehen mich auf
ihr Niveau herunter und schlagen mich dort mit ihrer Erfahrung! :p
(lt. alter usenet Weisheit) iPod, iPhone, iPad, iTunes, iRak, iDiot

On 11 Oct 23 at group /comp/lang/forth in article nnd$63d12f9f$31e9bfd0@3e1638bc22961201
<albert@cherry.> (albert@cherry.) wrote:

> In article <GQbfNFvEQoB@allinger-307049.user.uni-berlin>,
> Wolfgang Allinger <all2001@gmx.de> wrote:
> <SNIP>
>> He made a mistake while selling me some MC, one disk contains the complete
>> source for the MC Kernal. Poor documentation and quite a hack. I fiddled
>> several days to repair the bugs and recompiled 'my' MC-kernals.
>> So I could survive with my engineering office, delivering industrial
>> controllers and projects. :)

> Is their any value to this compiler? Should you preserve it for posterity?
> The product is no longer available, so you could publish it in Germany.

The last time I used LMI products on projects, was in the 90s.
Don't have the sources any more.

Later I used F-PC and Tcom by Tom Zimmer for realtime on PC-boards with my
own modified cooperative Multitasker with a crashproofed ABORT" and
QUIT... and a realtime clock (1ms) TSR (FINETIME) and IR-driven aWAIT...
This can't be done, said Tom :) I did it :))) Some 20 tasks running.

It was a big project with some 20 PC Stations in a production line,
working for more than 20yrs w/o any problems and killed SIEMENS and hp.

When I emigrated to PY, I sold all my PC and development stations
including SW to my customers. So I don't have any access any more.

Also used AmForth(?) (a no longer existing embilical TCL...
implementation) for cygnal F310 family.

> Groetjes Albert
> --
> Don't praise the day before the evening. One swallow doesn't make spring.
> You must not say "hey" before you have crossed the bridge. Don't sell the
> hide of the bear until you shot it. Better one bird in the hand than ten in
> the air. First gain is a cat spinning. - the Wise from Antrim -

Besser eine Taube im Bett, als eine Lahme auf dem Dach :)

Saludos (an alle Vernünftigen, Rest sh. sig)
Wolfgang

--
Ich bin in Paraguay lebender Trollallergiker :) reply Adresse gesetzt!
Ich diskutiere zukünftig weniger mit Idioten, denn sie ziehen mich auf
ihr Niveau herunter und schlagen mich dort mit ihrer Erfahrung! :p
(lt. alter usenet Weisheit) iPod, iPhone, iPad, iTunes, iRak, iDiot

On 11/10/2023 6:23 pm, Zbig wrote:
>
> I guess it was a stance of the type „priest of secret lore” and the
> desire to maintain (kind of) a monopoly? ;)

LMI got its start by offering Fig-Forth ready-to-run on CP/M using
OS files with tools and manual for $50 vs. $10 for a raw Fig listing.
It was very appealing. It was upgraded to Forth-83 and the price
increased to $100. Then came L&P F83. Much the same features - all
for free. Suddenly the bottom end of the market was gone.

> 400 USD even today it's still not that cheap. Yes, I'm aware it
> depends „what for” — but from what I see there were real problems
> to be solved/fixed.

Do you mean UR/Forth? Hugh couldn't praise it enough:
https://groups.google.com/g/comp.lang.forth/c/0IQudXjWoTw/m/BGbQEdK8AgAJ

On Wednesday, October 11, 2023 at 8:32:00 AM UTC+2, Zbig wrote:
[..]
> Hmm... so maybe I shouldn't bother Mr. Duncan after all.
I'm unpleasantly surprised that people here rather draw
conclusions based on a posting than check for themselves.

-marcel

> > Hmm... so maybe I shouldn't bother Mr. Duncan after all.
> I'm unpleasantly surprised that people here rather draw
> conclusions based on a posting than check for themselves.

You know: my interest in UR/Forth is purely histori/techni/cal,
so I'm not going to pay for it 400 USD.
OTOH I wouldn't to make impression on Mr. Duncan that I'm going
to rob him or something; that I wanted to extort anything from him.
Just was curious if he meanwhile (maybe) released (or would be
keen to) his Forth to the public, like to known place forth.org — even
without the sources.