FORTH for every Atari

[Moulinaie]

Guillaume, I LOVE what you have done here!

(I pretty much say this on all your stuff, I guess, because we seem to have several similar interests).

Such a phenomenal set of FORTH Words being created for the Atari range.

Looking forwards to the English manual. (So I must also thank Simon for his effort too).

Dont stop what you are doing!

Many thanks !
I was, some decades ago, a Parx developper and I kept the original docs.
But the sources have been lost and it's a pity because the system could be completed with new formats.
I've been in contact with two of the main programmers, the chances to recover the original work are close to zero...!

But, as it is, the system is usable and brings real facilities to manage images.
That gives me some ideas to modify M_PLAYER one day...

I hope the English manual will be soon available!

Guillaume.

[Moulinaie]

Hello,

The English manual has arrived!
Simon worked a lot on it but... while he was working, I made tons of changes in FORTH. I can't help myself...
So, when I received his version, there was still some chapters to add and lots of modifications. That's why it took some time...

What's new?

VERSION 0.3.0 includes the M&E modules management:

• read images

• save images

• dithering, zooms, effects

Pierre-Louis Lamballais and Eric Da Cunha gave me permission to publish a ZIP with the modules they created, you'll find it on my page.

The French manual has been updated

The English manual is available

Everything is on my page:

france2.gif

Page française: https://gtello.pagesperso-orange.fr/forth_f.htm

greatbr2.gif

English page: https://gtello.pagesperso-orange.fr/forth_e.htm

zip.gif

I merged several ZIP files and now a single archive contains:

• French an English versions

• 68030 and 68000 versions

• assembly source code

Guillaume.

[mrbombermillzy]

Hello,

The English manual has arrived!
Simon worked a lot on it but... while he was working, I made tons of changes in FORTH. I can't help myself...

You are a perfectionist, eh?

Great effort there. I for one am very grateful. I dont have time to learn GEM/AES coding for my projects, so I expect I will use this and it may turn out to be the stepping stone to greater things.

A while back, I was speaking to Fredrik Cordier, (who runs the Unity / AMOS project) and he showed me a way to use DEVPAC MACROs to build language keywords with arguments and constructs (like e.g. C), whilst circumventing the MACROs biggest flaw of compiling the variable/array data and the MACROs data on different passes, which results in the MACRO input data not actually being defined yet.

Unfortunately, Im not sure it can be used (or at least my comprehension of it) for multi dimensional arrays, which I need to process lots of, so maybe I can use your implementation of FORTH as a sort of 'super MACRO language' to create a BlitzBasic wrapper for the Atari line.

This would drastically reduce the porting time of my game to the Atari 68xxx systems, provided the FORTH language is fairly low level in its workings.

Anyway, thats eough of my drivel...congratulations on a job well done!

[Moulinaie]

Unfortunately, Im not sure it can be used (or at least my comprehension of it) for multi dimensional arrays, which I need to process lots of, so maybe I can use your implementation of FORTH as a sort of 'super MACRO language' to create a BlitzBasic wrapper for the Atari line.

Many thanks!

FORTH has not multidimensional arrays in standard, but that can be programmed.
How "multi" would it be ? Two dimensions? Three ?

I could add the feature..

Guillaume.

[mrbombermillzy]

FORTH has not multidimensional arrays in standard, but that can be programmed.
How "multi" would it be ? Two dimensions? Three ?

I could add the feature..

Guillaume.

Well, I was not expecting you to program this in at all as its not quite that simple....

The arrays I use can have up to (a few of them are at least) 4 dimensions in the syntatical format: Array1(Var1,Var2,Var3,Var4)

However, it gets more complex when I read in array data in 'tiers' and use an array from (e.g.) tier 1 in tier 2 arrays. So for example:

Array2(Var5,Var6,Var7,Array1(Var1,Var2,Var3,Var4)) or even all dimensions being arrays as is often the case.

I did start sucessfully converting the data, however, it was only possible to a bracketless formatting with MACROs on Devpac 3 Amiga with my Vampire V2, but its a lot of awkward format conversions, especially when a previously read in array is used inside another array as described above, which then become syntatically incorrect (too many arguments) without further processing.

I was wondering if you thought it at all possible that the formatting could be kept (as above) or very similar to this, with the multi dimensional arrays AND arrays within arrays as descibed above staying intact?

There would have to be some process which differentiates between arrays and variables within the presented data respectively. Quite a complex task I would imagine, so like I said, beyond what I could ever expect you to 'program in'.

Maybe there is a way to have the data set and ready for use in 'tiers' like this with FORTH and it isnt such a problem?

Anyway, sorry if I have gone off topic with this.

[viking272]

The functionality within this version of FORTH is amazing. I checked other versions on other platforms (when translating) and many features weren't there! Guillaume has done an amazing job and a labour of love over the last few decades for sure.
Everyone needs to download this and have a play!
Simon

[Moulinaie]

However, it gets more complex when I read in array data in 'tiers' and use an array from (e.g.) tier 1 in tier 2 arrays. So for example:

Array2(Var5,Var6,Var7,Array1(Var1,Var2,Var3,Var4)) or even all dimensions being arrays as is often the case.

Oh ! what a simple thing...

If your arrays have fixed dimensions, this can be done easely.
Do your index start at zero or 1 ?

Supposing they start at zero and that the dimensions of Array1 are (3,5,10,4), then it contains 3×5×10×4 = 600 elements
So you can declare:

Code: Select all

600 array ARRAY1
or
3 5 10 4 * * * array ARRAY1

Array1(x,y,z,t) is the element number:
x + 3 × (y + 5 × ( z + 10 × t ) )
So you can program a word ARR1ADR

Code: Select all

: ARR1ADR
   10 * + 5 * + 3 * + ARRAY1
;

That you'll use this way:

Code: Select all

x y z t ARR1ADR			\ this gives you the address of the element
x y z t ARR1ADR @		\ recalls on the stack the value of this element
value x y z t ARR1ADR !    	\ stores the value in the element

You can put parenthesis (they are ignored, just here for readability, but with spaces around!)

Code: Select all

( x y z t ) ARR1ADR		\ this gives you the address of the element
( x y z t ) ARR1ADR @		\ recalls on the stack the value of this element
value ( x y z t ) ARR1ADR !    	\ stores the value in the element

Something like that...

Guillaume.

[mrbombermillzy]

However, it gets more complex when I read in array data in 'tiers' and use an array from (e.g.) tier 1 in tier 2 arrays. So for example:

Array2(Var5,Var6,Var7,Array1(Var1,Var2,Var3,Var4)) or even all dimensions being arrays as is often the case.

Oh ! what a simple thing...

If your arrays have fixed dimensions, this can be done easely.
Do your index start at zero or 1 ?

Supposing they start at zero and that the dimensions of Array1 are (3,5,10,4), then it contains 3×5×10×4 = 600 elements
So you can declare:
.......

Something like that...

Well, thats the complication; the array dimesions are fixed, but the number of elements in each dimension are unknown, hence the having to read in each section (or I call them 'tiers' above) of data to get the sizes before using them on the next set (and so on).

So in my example above, the values of Var1-7 are unknown, so are loaded in, which has now enabled Array1 to be defined as well as the elements of Array2 to have a set value. THEN after this Array2 can be loaded in and defined, as the no. of elements are now known.

Hope this makes sense?

[Moulinaie]

Well, thats the complication; the array dimesions are fixed, but the number of elements in each dimension are unknown, hence the having to read in each section (or I call them 'tiers' above) of data to get the sizes before using them on the next set (and so on).

So in my example above, the values of Var1-7 are unknown, so are loaded in, which has now enabled Array1 to be defined as well as the elements of Array2 to have a set value. THEN after this Array2 can be loaded in and defined, as the no. of elements are now known.

Hope this makes sense?

You can modify the word ARR1ADR using variables as dimensions instead of constants:

Code: Select all

: ARR1ADR
   DIM3 @ * + DIM2 @ * + DIM1 @ * + ARRAY1
;

It's just a bit slower...

Guillaume.

[Moulinaie]

Hi,

I've released version 0.3.1 with:
- autoconfiguring of the TRM module to adapt it to every screen resolution/planes.
- AUTOEXEC.FOR can be run at start

I have added another example of a "Universal Image Viewer" in a very short source code.
It loads all the RIM modules (read image) and opens a fileselector. You pick up an image and, if the format matches one module, it is displayed from monochrome to 32 bits.

forth_mod3c.jpg

An example with a TIF file.

You'll find the commented source (in light blue at then end of the page) here:

(english) https://gtello.pagesperso-orange.fr/forth_e.htm
(français) https://gtello.pagesperso-orange.fr/forth_f.htm

Guillaume.

[mrbombermillzy]

Hey Guillaume, thank you for your help so far. I dont want to completely derail your thread, so will attempt to use the words in the way you have described, if I can parse the input file arrays into an appropriate order. Maybe I will PM you instead with any further questions, if thats ok with you?

[Moulinaie]

Hey Guillaume, thank you for your help so far. I dont want to completely derail your thread, so will attempt to use the words in the way you have described, if I can parse the input file arrays into an appropriate order. Maybe I will PM you instead with any further questions, if thats ok with you?

That's fine for me !

Guillaume.

[JeanMars]

Amazing work !

Thanks for your efforts Guillaume!

Jean

[Moulinaie]

Amazing work !

Thanks for your efforts Guillaume!

Jean

Merci Jean !!!
(sorry for the delay, I'm recovering from COVID...)

Those Atari are so nice to program.
Guillaume.

[DarkLord]

Ugh, hang in there and get better soon!

[farvardin]

hello @Moulinaie

now that I've a correct setup for using Forth for Atari ST, I've tested it a bit further.

I'm in no way a forth programmer, I've just begun to learn it. I've noticed a few things which are different from the ANS forth I've encountered. I'm in no way saying your forth is doing it wrong, you seems to be a real forth wizard, but I've just wanted to stress out that some behaviors can be confusing for newcomers:

1/ the .s command is giving the stack number in the reverse order of the one in ANS forth (for example pforth or gforth):
1 2 3 4 5
will result in
5 4 3 2 1 with the .s command.

In Starting Forth the author writes "Remember, the rightmost represents the topmost.". https://www.forth.com/starting-forth/2- ... rithmetic/

2/ after a word was defined, if we define the same word again, in ANS forth there is only a warning (for example: "HELLO already defined"), but the new definition replace the older one. Then, if we "forget HELLO", the new definition will be replaced by the older one. In Atari Forth, we cannot override the old definition, we have to "forget" it before.


Probably there are good historical reasons for this! It looks like you're working on this Forth for a long time, maybe even before ANS forth was normalized?

[mOdmate]

Btw, Escape released a demo called "Monscape" for the Atari TT in July that was coded in Forth. Source code is included in the package.

https://demozoo.org/productions/310231/

[Moulinaie]

hello @Moulinaie

now that I've a correct setup for using Forth for Atari ST, I've tested it a bit further.

I'm in no way a forth programmer, I've just begun to learn it. I've noticed a few things which are different from the ANS forth I've encountered. I'm in no way saying your forth is doing it wrong, you seems to be a real forth wizard, but I've just wanted to stress out that some behaviors can be confusing for newcomers:


2/ after a word was defined, if we define the same word again, in ANS forth there is only a warning (for example: "HELLO already defined"), but the new definition replace the older one. Then, if we "forget HELLO", the new definition will be replaced by the older one. In Atari Forth, we cannot override the old definition, we have to "forget" it before.

Some historical points:
I started programming FORTH long time ago and at the time, I only had the user manual of a FORTH that ran on the Thomson TO7 (a french 8 bits computer). I never read the "standard" and went on programming my own version. That's why differences exist! I don't know when ANS Forth was normalized, do you?

I can modify the .s command, that will be easy to do and will conform to the ANS Forth.

Before, my FORTH accepted to replace an old definition with a new one. But, the memory used by the old one was not freed! And even after a forget, only the last definition was really removed from memory, the old was still using bytes but not accessible!
So, I preferred to output an error message in this case. It would require some extra work to chain the definitions and go back to the old one. And... I don't know if it's worth it and if it's useful.

Thanks for your remarks !!!

Guillaume.

[farvardin]

> I don't know when ANS Forth was normalized, do you?

It seems to have been in 1994:

Common practice was codified in the de facto standards FORTH-79] and FORTH-83 in the years 1979 and 1983, respectively. These standards were unified by ANSI in 1994, commonly referred to as ANS Forth.

But it's more complicated than that:

What ANS Forths are available ?
The simple answer is: none

http://www.forth.org/ansforth/ansforth.html

And it's true there a several different Forth implementations, as if every programmer needed to custom his own version.

I suppose you had this TO7 manual?
http://dcmoto.free.fr/documentation/for ... index.html

This manual, like many TO7 manuals (including the BASIC ones), are very clearly written. Much more than the official Forth docs, which are confused and lack examples.

It must have been an incredible amount of work to design your Forth from this manual, did you have access to the TO7 forth software as well?

I've tested this TO7 forth at the beginning of this year (I've used TO7 in my youth, but not with FORTH), and I enjoyed very much the fact that you have a direct access to the machine, including graphics and sounds. It's the problem with "modern" forth, most of them are console-only. You can create some graphics with gforth but it's more complicated. I'm glad the ST Forth can be graphical tool. I haven't seen anything related to the YM2149 chip for sounds in ST Forth, is it planned in the future?

I can modify the .s command, that will be easy to do and will conform to the ANS Forth.

I've noticed the .s command in the TO7 forth is like in ANS forth:



(you can see this command in http://dcmoto.free.fr/documentation/ini ... index.html too)

I suppose it would make sense to modify this .s command in ST forth, if that makes sense for you too!

I've made a little vector image with your forth, at the moment I can't do much more:


I don't really know how to make a little animation (like a bouncing ball), probably it should be made with interruptions, but it's probably above my programming skills...

[Moulinaie]

> I don't know when ANS Forth was normalized, do you?

It seems to have been in 1994

I started in 1989 I think. Learning both 68k assembler (to program on Atari) and FORTH (to know what to program!).

I suppose you had this TO7 manual?
http://dcmoto.free.fr/documentation/for ... index.html

This manual, like many TO7 manuals (including the BASIC ones), are very clearly written. Much more than the official Forth docs, which are confused and lack examples.

It must have been an incredible amount of work to design your Forth from this manual, did you have access to the TO7 forth software as well?

Yes that manual that I couldn't read for so long as it belonged to my university's library!
No, I didn't have the Thomson to test it, just the manual and my own version.

I'm glad the ST Forth can be graphical tool. I haven't seen anything related to the YM2149 chip for sounds in ST Forth, is it planned in the future?

What would you like to do with the Yamaha?
If it's the standard use (activating one voice with frequency and volume), isn't there DOSOUND in the Xbios?
If it's to replay digitized sounds (as the DMA), I have the routines in MP_STE but they require some extra tables and is it worth it now that the STE/Falcon/TT have the DMA system ?

I've made a little vector image with your forth, at the moment I can't do much more:


I don't really know how to make a little animation (like a bouncing ball), probably it should be made with interruptions, but it's probably above my programming skills...

Cool work !
A bouncing ball requires, for example, the use of vro_cpyfm to blit an image on the screen.
Once there was the LineA that had some basic "sprite" instructions (used for the mouse I think).

Guillaume.

[Moulinaie]

1/ the .s command is giving the stack number in the reverse order of the one in ANS forth (for example pforth or gforth):
1 2 3 4 5
will result in
5 4 3 2 1 with the .s command.

Hello !

You'll find on my page the not finished version 0.5.3 with ".s" following the standard.
(The top of the stack is on the right)
If, on the flag line of FORTH.INF you add a "S", then my old routine is used to override the standard,
(The top of the stack is on the left, this is cool for 8-bytes floating point values that are split into two longs, they appear in the right way)
If you don't put the S, then the default is the standard.

Guillaume.

[farvardin]

salut,

that's great, I've upgraded my version to the latest You did well to keep the old behavior as an option.

About the sound, I just wanted to be able to drive the PSG (making some beeps and noise). I wasn't aware of the xbios stuff. With dosound I was able to make a fading noise with this code:

Code: Select all

0 80 1 3 8 dosound 

But I haven't find how to make it sounding more interesting.

[Moulinaie]

salut,

that's great, I've upgraded my version to the latest You did well to keep the old behavior as an option.

About the sound, I just wanted to be able to drive the PSG (making some beeps and noise). I wasn't aware of the xbios stuff. With dosound I was able to make a fading noise with this code:

Code: Select all

0 80 1 3 8 dosound 

But I haven't find how to make it sounding more interesting.

Argh! You can't send the values this way, they must be stored as a string in memory,a nd only the string address is passed to dosound.

Example:

Code: Select all

here                  \ put first free address on the stack, this address will be used by dosound

remember			 \ remember the current state of memory

\ deux notes

 %h08000900 ,            \ fills memory with hex values ',' for a long
 %h0a00001c ,
 %h0101073E ,
 %h080f8211 ,
 %h009c0102 ,
 %h82110800 ,
 %hff00 w,			\ and "w," for a short

dosound			\ treat sound with "here" as address
restore                      \ restore memory state 

This outputs two notes on canal A.
If you want to do the bell sound, here are the bytes (taken from the ROM):

Code: Select all

\ cloche

 %h00340100 ,
 %h02000300 ,
 %h04000500 ,
 %h060007fe ,
 %h08100900 ,
 %h0a000b00 ,
 %h0c100d09 ,
 %hff00 w,

For the keyboard clic :

Code: Select all

\ clic clavier

%h003b0100 ,
%h02000300 ,
%h04000500 ,
%h060007fe ,
%h08100d03 ,
%h0b800c01 ,
%hff00 ,

If someone has better examples... I think that whole files with DOSOUND codes existed and could be played easely.

Guillaume.

[mrbombermillzy]

On a (hopefully not too far removed) similar topic...

When I do get the chance to go there, I was very much looking forwards to making use of the support for the Replay16 cart. However, the manual states that this is discontinued. Has this support been (or will be in future builds) removed?

Again, many thanks for your continued work on this amazing language.

[Moulinaie]

On a (hopefully not too far removed) similar topic...

When I do get the chance to go there, I was very much looking forwards to making use of the support for the Replay16 cart. However, the manual states that this is discontinued. Has this support been (or will be in future builds) removed?

Again, many thanks for your continued work on this amazing language.

Hi!
Thanks for your interest.

The former grind of the STReplay16 was made when FORTH included an internal assembler. Some features of the STReplay support relied on this assembler. As this one has disappeared ...

The set_replay & replay instructions still exist and manage the TimerA to run a little interruption routine that you have to write now by yourself.
I don't know what you wanted to do with your cartridge.
If it's a simple replay (sample in memory), a listen (take from in to out) or record (in to memory), I can imagine standard little routines for that and you won't have to use the external assembler.

Guillaume.