New French website dedicaced to Pure Pascal

[neotenien]

Hello

In 2021 May, I put on line a new french website dedicaced to Pure Pascal.

Actually this website contains:

• Main page explaining all about pascal language, what is the properties of Pure Pascal, where to download version 1.1 and 1.1.1 version... And some other Pascal compilator in 8 bits world
• "Lexique" Which is the translation from German of the "PASCAL.HLP" file and converted to french (it also contains the Pchar type about version 1.1.1). It's the dictionnary of Pure Pascal Language (And Turbo Pascal 7+). 75 Pages from HLP files
• "Unités" is the part of all units from Pure Pascal, translated from UNIT.HLP (german) and converted into HTML. There are 485 pages from HLP translated and converted in HTML here.
• "Algorithmes" : here are some personnaly functions and procedures I added and will add to complete Pure Pascal for being close to Free Pascal features (and even more). Actually, I only added 41 extended Maths functions with 9 that doesn't exist on Free Pascal!

Pure Pascal seems really to be a very very good compiler for our Atari computers. It is said that it can compile 1000 lines/s on normal Atari Falcon... Many faster that any C compiler.

Thanks to Google translator I could translate all German HELP file in French. It take me some times but this job need to be done. Not all is done yet (Some Unit are not translated yet, and it remains all the IDE Pascal to translate).

Of course all of this has the goal to make developpers handle Pascal language and appreciate it as it deserve. All GEM/TOS fonction that exist in Pure C, Lattice C, also exist in Pure Pascal.

There are some specific Pascal stuff as set type (i don't know any other language where it exist), limited selection, polymorphisme (as read, write, the "+" operatoir etc) that doesn't exist in C. And also, the flavor in Turbo PAscal/Pure PAscal are the CRT and GRAPH Unit which doesn't exist on C language.

Some very good and advanced softwares are compilmed in Free Pascal Lazarus. And as I know, some software in ASH were compiled in Pure Pascal (but I don't know which one).

Have fun.

[vinz6751]

Bravo, super initiative !

[neotenien]

Bravo, super initiative !

Merci Vincent R.!

Dis moi pendant que je t'ai sous la main (hors sujet)=... Est ce qu'il se vend toujours des FireBee chez Medusa System ?

Et autre question, j'ai vu parmi une de tes pages web que tu avais mis des liens vers fVDI mais ce sont des liens obsolète et actuellement, je n'ai toruvé aucune ressource pour le télécharger (format ZIP). Ya juste un truc sur Githhub mais github, pas de lien pour télécharger.

Concernant Pure Pascal, ya de quoi faire des truc vraiment super avec les Unité GRAPH et CRT. Ya juste qu'il faut que je vois comment est le format Pascal des images (Avec GetImage et Put Image) et que c'est limité, hélas, à 16 couleurs. Mais une fois que j'aurai trouvé comment est ce format, créer des jeux vidéo 2D sera vraiment très simple.

Et si je trouve des Unité de conversion d'image en Pascal (dont le GIF!!) ça m'arrangerait bien finalement. Le logiciel LazPaint (Lazarus paint) est vraiment un bon logiciel même s'il est loin d'avoir des fonctions étendus (grille de Bézier existante, ou des effets qu'on trouve dans certains logiciels comme Smurf, par exemple) qu'on trouve dans certains logiciels Atari.

[vinz6751]

I'm Vincent B, not Vincent R.

Prepare to be told off for speaking French on an English-speaking forum

The Free Pascal is looking for an Atari TOS maintainer, isn't it something you would be interested in doing ?

AFAIK the FireBee has been completely halted for several years. 2 years ago I asked about it but there was no new manufacturing planned.
I'm not aware of other Coldfire-based projects, but there are many 68k projects in existence.
For example, the H5 (remake of STf board), or V4SA (Olivier Landemarre is very active there) which seems to emulate some Atari hardware not too bad. I'm personally involved in porting EmuTOS on the Foenix Retro Systems 68k-based computers, like the A2560 and GenX.

fVDI has been maintained under the "FreeMiNT" umbrella (like XaAES), you can find it here https://github.com/freemint/fvdi
I'm actually trying to make a VDI for the Foenix A2560U so if the VDI is something of interest to you I'd be willing to chat about it !

[mfro]

... Ya juste un truc sur Githhub mais github, pas de lien pour télécharger...

fVDI snapshots (bleeding edge CI-generated from latest github sources) can be downloaded here: https://tho-otto.de/snapshots/fvdi/

[Moulinaie]

Hello

In 2021 May, I put on line a new french website dedicaced to Pure Pascal.

Actually this website contains:

• Main page explaining all about pascal language, what is the properties of Pure Pascal, where to download version 1.1 and 1.1.1 version... And some other Pascal compilator in 8 bits world
• "Lexique" Which is the translation from German of the "PASCAL.HLP" file and converted to french (it also contains the Pchar type about version 1.1.1). It's the dictionnary of Pure Pascal Language (And Turbo Pascal 7+). 75 Pages from HLP files
• "Unités" is the part of all units from Pure Pascal, translated from UNIT.HLP (german) and converted into HTML. There are 485 pages from HLP translated and converted in HTML here.
• "Algorithmes" : here are some personnaly functions and procedures I added and will add to complete Pure Pascal for being close to Free Pascal features (and even more). Actually, I only added 41 extended Maths functions with 9 that doesn't exist on Free Pascal!

Hi !
Great ressource !
Is it possible to increase the size of the main menu in the top left corner? As it is followed by a large advertising, I didn't see it... It's only when I read completely your message that I knew there was different sections and tried to find them!

Guillaume.

[neotenien]

Hi !
Great ressource !
Is it possible to increase the size of the main menu in the top left corner? As it is followed by a large advertising, I didn't see it... It's only when I read completely your message that I knew there was different sections and tried to find them!

Guillaume.

Hello Guillaume

Thanks for you message

Of course it can be done! Either in HTML (with FONT tag not 100% compatible www but it wirk with all Atari Webroiwsers) either using CSS (Maybe it can wotk with HighWire). Of course these siez and background color are fully arbitrary! I am not at all an artist LOL. I'll do it tomorrow, now ot's time of "Kuroko's basket" on Game One.

[neotenien]

... Ya juste un truc sur Githhub mais github, pas de lien pour télécharger...

fVDI snapshots (bleeding edge CI-generated from latest github sources) can be downloaded here: https://tho-otto.de/snapshots/fvdi/

Thank you very much!!

Why does all these old software FTP source (as lip6, funet etc) are not mainained anymore ? They always exist but not with update since maybe ... 20 years ago. Before we found all rsc in these servers...

[neotenien]

I'm Vincent B, not Vincent R.

The Free Pascal is looking for an Atari TOS maintainer, isn't it something you would be interested in doing ?

Not at ALL!! Pure Pascal is very suffisiant fot even modern software! As I told difference with Delphi are very few. And I don't have skulls about compiling compilers from sources.

AFAIK the FireBee has been completely halted for several years. 2 years ago I asked about it but there was no new manufacturing planned.

I'm not aware of other Coldfire-based projects, but there are many 68k projects in existence.
For example, the H5 (remake of STf board), or V4SA (Olivier Landemarre is very active there) which seems to emulate some Atari hardware not too bad. I'm personally involved in porting EmuTOS on the Foenix Retro Systems 68k-based computers, like the A2560 and GenX.

In the english pages of firebee, there are recent news (December 2022), About a very new FreeMiNT setup and about a new version of EmuTOS. FireBee is not down at all!! There are many people workig on it, and the website was out because of the update of the CMS used (I think it's WordPress but not sure). This is why there were no news since Covid crisis. .But it's gone now! The website work now.

Sorry about the Vampire card you talk about but for me the FireBee computer is many more advanced and interesting computer!! Using a RISC processor as the CF, it's an low energy processor, and it's 400 MIPS in term of power (100 time more than a basic Falcon and almost 400 times with the Atari ST), and it's a micro controler too, allowing Ethernet, USB etc...
.

[mfro]

Sorry about the Vampire card you talk about but for me the FireBee computer is many more advanced and interesting computer!! Using a RISC processor as the CF, it's an low energy processor, and it's 400 MIPS in term of power (100 time more than a basic Falcon and almost 400 times with the Atari ST), and it's a micro controler too, allowing Ethernet, USB etc...
.

You appear to be a FireBee fan? I'm as well.

If that's the case, I do not understand why you would target your efforts to PurePascal. PurePascal for sure is a nice piece of software, but it's unable to generate ColdFire native code. That means your PurePascal programs will only run with severely reduced performance on the FireBee.
While the 100x performance you mentioned for the v4e compared to the Falcon is more like 80x in reality with native ColdFire code (see czietz' CoreMark page), it's more like 40 to 50x for less compute-intensive, real world programs (that need to do regular screen updates, for example).

The 68k emulation (done using MicroAPL's cf68klib) required for PurePascal code will reduce performance severely as each emulated instruction needs to go through the cf68klib's exception handler that replaces unavailable m68k instructions with equivalent ColdFire code sequences.
Your PurePascal code will end up (depending on its nature, i.e. the number of instructions that need to be emulated) rather somewhere near 5-10x Falcon speed than 100x, while FreePascal (native ColdFire compiled) code will run somewhere near 10x faster.

Long story short, if your efforts are going to target the FireBee, your programs would be a lot more effective with FreePascal than PurePascal.
IMHO, it would be a shame to waste the FireBee's true power in emulation for new code.

[neotenien]

You appear to be a FireBee fan? I'm as well.

Yes

If that's the case, I do not understand why you would target your efforts to PurePascal. PurePascal for sure is a nice piece of software, but it's unable to generate ColdFire native code. That means your PurePascal programs will only run with severely reduced performance on the FireBee.
While the 100x performance you mentioned for the v4e compared to the Falcon is more like 80x in reality with native ColdFire code (see czietz' CoreMark page), it's more like 40 to 50x for less compute-intensive, real world programs (that need to do regular screen updates, for example).

The 68k emulation (done using MicroAPL's cf68klib) required for PurePascal code will reduce performance severely as each emulated instruction needs to go through the cf68klib's exception handler that replaces unavailable m68k instructions with equivalent ColdFire code sequences.
Your PurePascal code will end up (depending on its nature, i.e. the number of instructions that need to be emulated) rather somewhere near 5-10x Falcon speed than 100x, while FreePascal (native ColdFire compiled) code will run somewhere near 10x faster.

About PP choice:
1/ Actually I don't have any FireBee
2/ Lazarus IDE is not easy to use
3/ My goal is to produce a Pascal Code portable with all version of Pascal (even Free Pascal), knowing that the minimal base should be TP 5.5.

About CF Performance, i suggest you this article on FireBee Website. What we learn here:
1/ Atari Falcon is only 1.5 MIPS whereas it should be 3.8.
2/ The ST RAM decrease performs (especially with TT and FireTOS).
3/ Comparing bewteen 68060/CF, with same frequency (266/60 Mhz = 4), we see that even here, CF compiled MIPS Drystone is better than 68060 specific (Even if we take the best Dhrystone on 68060, the 68020 compilation as 71 MIPS) : we have 303 to 312 MIPS for the CF... So 303/71 = 4.27 times (more than the 4 report seing frequency) and 312/71=4.4 times better. So CF processing using TT Ram (or even ST Ram on Emutos) is better than 68060.
4/ Comparing CF with Falcon 030 : It's 22.6 times better than Falcon if compiled as 68000 and using ST Ram, 28.7 times better than Falcon, using 68k compiling and TT Ram, and above all, more than 200 time faster than Falcon 030 if using CF compilation and TT Ram. So a TOS sofwware should work 22 times faster at last than a Falcon (using ST RAM).
5/ We see that despite it is announced than CF should be 400 MIPS it's here 312 max only... Why ? Due to TT Ram ?

About rare not compatible code on 68k as it is said, they are rare. I wonder if PP produced by ASH, use this rare code... I think it depends of compilation option maybe ? Am ,not rather advanced in PP IDE.

About Free Pascal. Surely there are many valuable softwares existing on it (As Laz Painter for instance), so it could be interesting compiling in CF code. Actually I'm split in my choice. I use PP for train myself, try ti see the difference with FreePascal, add function and procedure for PP to have same units as Free Pascal... And the main problem is that I'm good at all easy with all GEM/AES Stuff. I hav to do some sofwtare on this environment. Well... C Language is mainly used on Atari Software but Pascal Language is many more easy and secure to code.

Well I also watch some video with FireBee under FreeMiNT... When I see all these GEM Game which were very slow on Falcon 030 (1 to 2 FPS) and very fast (normal speed) on Firebee, this let me think that FireBee is a very good alternate computer for the PC one. Even if, actually, only 400 MIPS versus 4 GMIPS for PC at least...

[mikro]

i suggest you this article on FireBee Website. What we learn here:

I can assure you that Markus (mfro) really doesn't need any lecturing on FireBee performance - he is an owner, user and active developer on the actual machine for several years now (in stark contrast to you).

Also, I would advise to pay more attention to his reasoning, it seems that you're missing the point about emulated environment all together, so your 'performance analysis' is really off because it doesn't apply to your Pure Pascal choice.

[neotenien]

i suggest you this article on FireBee Website. What we learn here:

I can assure you that Markus (mfro) really doesn't need any lecturing on FireBee performance - he is an owner, user and active developer on the actual machine for several years now (in stark contrast to you)

The webpage I deliver here show true performance of all Atari system in erm of drystone, so it show the true performance of processor out of system.

Also, I would advise to pay more attention to his reasoning, it seems that you're missing the point about emulated environment all together, so your 'performance analysis' is really off because it doesn't apply to your Pure Pascal choice.

In the webpage, there are no environment consideration, it's purely in each computer (TT, Falcon, Falcon 060, FireBee cipiled in 68000 mode, CF mode etc)
And concerening what you told about emulate environment, I don't know what you talk about that, that's not the subject at all!! I'm computer coder since 1990 (began with Pascal, then, some rare stuff in C, then PHP/JavAScript). If you talk about the rare 68000 instruction not present on the coldfire, hat's not a problem at al, software using that may be very rare so we should drop that softwares.
And about my choice for Pure Pascal, it's a pertinent choice as I don't have any FireBee, and Pure Pascal has sifficiant advanced functions/procedures/instruction to build major need. And even, maybe it would be a good sruff to build an unit "GEM2GTK" to make software portable to Fre Pascal under GTK

Does it exists a GEM2GTK library in C language ?

The 1st Atari ST software I wan to build in Pure Pascal is a statistic basket take that I already did on JavaScript. But I wonder if GEM/AES can create scoll menu ?...

[AnthonyJ]

The webpage I deliver here show true performance of all Atari system in erm of drystone, so it show the true performance of processor out of system.

I think you need to read it more carefully, along with what you're being told by people who know more about the FB / CF than you. Clearly the CF native binary runs significantly faster than the non-CF binary.

And concerening what you told about emulate environment, I don't know what you talk about that, that's not the subject at all!!

You see, this is what you're missing. The CF simply cannot run standard 020-060 binaries at all if they contain the non-implemented instructions. To make them work, there is a library which emulates the missing instructions by adding an illegal instruction vector, and implementing the instruction, and then returning to the application as if the instruction hadn't caused an illegal instruction exception. Obviously this slows things down considerably as 1 instruction turns into an exception handler + multiple other instructions to implement the "missing" instruction. This is the difference between the 303 vs 43 results recorded against the Firebee in your link.

If you talk about the rare 68000 instruction not present on the coldfire, hat's not a problem at al, software using that may be very rare so we should drop that softwares.

So, I guess that you agree that the answer is "drop pure pascal" and use Free Pascal since it isn't CF aware.

Your assumption that they are "rare" is not bourne out by the benchmark you posted, nor the lived experience of the Firebee owners posting in this thread (not me) who are advocating you don't create more software targetting a compiler that doesn't support generating CF native binaries.

Perhaps read this to get a feel for the instructions that are missing - word/byte versions of most maths ops for example. Not really "rare", and the sort of thing people might naturally use when dealing with data-items of those sizes...

[neotenien]

Well

Outside the compiler stuff, that I don't care finally (and it's NOT my problem!!), the goal of my website is to provide many pascal source "compatible" with, at less, Turbo Pascal 5.5 (or 7+), Pure Pascal, HS Pascal, etc... Pure Pascal provide wide units for CRT, Graph, System, DOS, and also for all TOS/GEM STuff.

It's why I provide 41 function in TP7/Pure Pascal for mathematics to update TP7 to FreePascal libraries.. (adding few not existing one in Free Pascal, as 2 Fibonacci functions (one very fast uning Binet formula), Fraction functions, GCD etc).

I'll keep on on this way... adding "old" Pascal fonctions for update to free pascal and many new ones in many domain (actually, maths, statistics, but after, wordprocessor, image manipulations etc).

Well, sadly, Alice Pascal, which is freeware, is not update anymore (toi be TP7 compatible), and is more limited than Pure Pascal. My goal is to provide source foe every compiler ... Free Pascal may compile these file also... But the goal is that software can be use to all Atari TOS system if the software doesn't require many ressource (as for JPEG, MPEG, etc...) even if we can consider that almost all kind of software already exist in TOS system.

[neotenien]

The webpage I deliver here show true performance of all Atari system in erm of drystone, so it show the true performance of processor out of system.

I think you need to read it more carefully, along with what you're being told by people who know more about the FB / CF than you. Clearly the CF native binary runs significantly faster than the non-CF binary.

Are you stupid ? I Never said the opposite, what i present in Benchmark about FireBee exacltly show what you wrote and I never told the opposite...

And concerening what you told about emulate environment, I don't know what you talk about that, that's not the subject at all!!

You see, this is what you're missing. The CF simply cannot run standard 020-060 binaries at all if they contain the non-implemented instructions. To make them work, there is a library which emulates the missing instructions by adding an illegal instruction vector, and implementing the instruction, and then returning to the application as if the instruction hadn't caused an illegal instruction exception. Obviously this slows things down considerably as 1 instruction turns into an exception handler + multiple other instructions to implement the "missing" instruction. This is the difference between the 303 vs 43 results recorded against the Firebee in your link.

I know that and IT'S NOT MY PROBLEM, another time! Pure Pascal can of course comple with specific 030 instruction, 68881 or not and with many other compiling directives. I use it for test my Pascal algorithm and all is ok until now.

You didn't understand what my website is for! It to provide list of Pascal words,; tutorial, types, etc, all Unit descriptions, and provide function and procedure compatible with as much as pascal compiler as possible! And to do that, it's necessary to know little difference between last versions or Turbo pascal (Pure Pascal compiler is fully compatible with Turbo Pascal 7 that was used for many PC softwares).

If you talk about the rare 68000 instruction not present on the coldfire, hat's not a problem at al, software using that may be very rare so we should drop that softwares.

So, I guess that you agree that the answer is "drop pure pascal" and use Free Pascal since it isn't CF aware.

Another time IT's NOT my problem!! I don't care about that!

[ThorstenOtto]

You simply don't understand the problem. It's not only about 030 and FPU instructions. PP will (as almost all other compilers for Atari) also generate instructions like add.w d0,d1, and those simply do not exist on cfv4e, generate an illegal instruction exception and must be emulated by 68kemu.

[AnthonyJ]

If you talk about the rare 68000 instruction not present on the coldfire, hat's not a problem at al, software using that may be very rare so we should drop that softwares.

So, I guess that you agree that the answer is "drop pure pascal" and use Free Pascal since it isn't CF aware.

Another time IT's NOT my problem!! I don't care about that!

That’s fine, if you don’t care about things working on the firebee, that’s OK - eg ST game/demo coders don’t care about that either I guess. However:

But the goal is that software can be use to all Atari TOS system if the software doesn't require many ressource

It’s highly likely that most software compiled by Pure Pascal cannot be used on a Firebee running EmuTOS, and will not run as well as it should on a firebee running FireTOS.

Advocating for the use of Pure Pascal over Free Pascal is contrary to this goal.

[neotenien]

You simply don't understand the problem. It's not only about 030 and FPU instructions. PP will (as almost all other compilers for Atari) also generate instructions like add.w d0,d1, and those simply do not exist on cfv4e, generate an illegal instruction exception and must be emulated by 68kemu.

It's ok

I use PP to test Pascal code on Atari TOS... That doesn't mean it can't be compiled by FreePascal after. The goal of my website is to provide as portable Pascal code as possible in all environments (and tested mainly on Pure Pascal but not only). So that, on HighSPeed Pascal, Alice Pascal etc. And Free Pascal of course. Pure Pascal is Turbo Pascal 7+ compatible code compiler whereas Turbo Pascal in on Intel processors... Like GNU C for all platforms

This is why I wrote "I don"t care about ColdFire problems". Even if PP offer some few favors.

That's sad the PURE pascal source doesn't exist in order to compile it for ColfFire BUT maybe Alice Pascal can be done ? Alice Pascal exist for Atari ST and Intel.

And I just thought of one thing... Why not writing a Pascal compiler on Coldfire uning Pure Pascal itself ? At begining of Pascal I hav heard that some compilers were writing in Pascal! Pure Pascal can insert some ASS part inside his code...ASH did create a Pure Pascal compiler, tey weren't a lot of people doing that and they include option to include 68020/68030 code, FPU or not etc... Why this would not be possible actually whereas "adapting" a free pascal compiler which, finally, dont have so much more (even for Dynamyc array, I found a solution to build this on standard Pascal)... Writing a Pascal syntax analyser and doing some ASS routine to compile it in CF using Pure Pascal, that's may not be so difficult ?

[ThorstenOtto]

ASH did create a Pure Pascal compiler, tey weren't a lot of people doing that and they include option to include 68020/68030 code, FPU or not etc...

But they had the source. They didn't write everything from scratch.

Writing a Pascal syntax analyser and doing some ASS routine to compile it in CF using Pure Pascal, that's may not be so difficult ?

You can try of course, but i doubt that this can easily be done.