Fortran 77 on Atari 8 bit?

[JamesD]

"Why there was no business oriented 6502 OS/machine or why nobody ported Fortran anyway?"

That isn't exactly what I would call clearly.

 

Maybe if you only read that one line.

 

The part you conveniently skipped:

 

You have to remember that CP/M started on the 8080 and Flex started on the 6800, both of which predate the Apple, TRS-80, and PET.

They only used terminals or teletype machines for user I/O and even use paper tape for data storage.

Being on the bleeding edge of personal computing people saw the potential to use computers for business and scientific purposes.

Once these machines had a foothold in the business and science worlds it was inevitable common tools would become available.

The funny thing is, Fortran probably wasn't ported until after 6502 machines like the Apple II and PET were out there.

I'm pretty sure fortran was out for Flex by 1980 though.

...

FWIW, some people did try to release a 6502 based computer along the lines of CP/M and Flex systems.

I believe the single board computers were so much cheaper none of the computer stores picked them up.

I don't think the manufacturer had a working OS yet either.

 

The KIM 1 had the potential to get a universal 6502 CP/M like OS.

I vaguely remember a lot of expansions being advertised for it but Commodore killed it off and offered the PET not long after purchasing MOS.

There was a lot of stuff in MICRO on the KIM 1 and I wouldn't be surprised if some people had released some sort of OS for it.

Once the KIM was dead so was pretty much anything major like that for it.

I guess the 6502 just didn't have enough lead time to develop it's own portable OS before the "trinity" of personal computers hit the market.

...

 

All I talked about was CP/M, Flex, the KIM 1 single board 6502 machine, the "trinity" of personal computers... things happening in the 70s and all 8 bit machines. The last date I mention was "by 1980". So if you read the entire message instead of just the first line, I'd say it's pretty clear.

 

Business isn't just 700 employee corporations. If you read some of the history of the early personal computer market you'll find dealers were selling a lot of CP/M and Flex systems. I think this came up in a discussion about the SOL. Even if they only sold 100,000 SOLs as I think the discussion mentioned, that's more total computers than IBM had manufactured up to that point by a wide margin. There was clearly some demand for the machines. Wasn't it IBM that predicted there was a market for only 5 such machines

 

My point was that the 6502 never developed a portable OS standard similar to CP/M so it never developed a similar business or scientific application market to draw software from for later machines. Had there been some 6502 market like that it probably would have developed a similar market and many of the same apps/tools. It was a cheaper CPU so it should have caught on.

 

But as I later stated, the Apple II did have a Fortran compiler. It was sort of the business and scientific machine of the 6502 world but everything was proprietary so most of the business or scientific stuff never made it to the Atari or other 6502 machines. I will say it's entirely possible there is another 6502 Fortran somewhere out there but I wasn't able to find it.

C has pretty much replaced Fortran and Pascal for everything these days so I don't see much demand for either language at this point.

 

The CP/M computers that did make it as a minor hit like Osborne and Kaypro, pretty much grenaded as corporations. They were exactly the reason why corporations would not buy any computer before IBM got in the market.

 

Again... you are talking about major corporations.

 

And Osborn & Kaypro were far from the only CP/M machine manufacturers. Altair, IMSAI, Northstar, SOL, Televideo, Tandy (Model II & Model IV), etc... all suppored CP/M.

A mom & pop store isn't going to buy an AS400 and the PC didn't exist in the period I was talking about.

 

You can argue with me all you want about your point of view but it's not what I was talking about at all.

 

[ricortes]

I'm not sure your reading what I am saying either. You are moving the wide spread adoption of personal computers back to the 70s let alone business use. It's like saying the hippie days were the 50s or Disco was the 60s.

 

I worked in Pharmaceutical R&D. Everyone I worked with a pretty much a an engineer, chemist, physicist, or some flavor of science major. Our company was headquartered in the Stanford Industrial Park. Next to nobody i.e. three people out of 700 had personal computers through the 70s and those were ones we built ourselves.

 

Look at the work Curt Vendel has done retrieving information about Atari Corp. Most of what is of interest he had to retrieve off of tape backups from mid/mainframes and this is true into the 80s. I'm not sure when Atari started using personal computers but I have yet to hear someone picked up an Osborne computer with Atari's payroll or a sales database on it.

 

*BUT* this would be a great time to get it resolved if the interview with Nolan hasn't happened yet. Just add it to the questions. I think Atari had something like 7,000 employees when it tanked. I'd be interested to know just how many pre 1980 personal computers were in use there minus the 400/800 of course.

[JamesD]

There have been several threads on sales numbers.
Even ultra rare systems like the Sphere 1 supposedly sold over 1000 machines and the Altair sold thousands in the first month alone.
It seems to me that there were 400+ thousand TRS-80s sold in 1979 alone and Radio Shack was selling them as fast as they could build them.
I would guess based on memory of several threads that total numbers were at least 2 to 3 million once you group all the 8 bits together.
Out of that it would not be unreasonable to assume thousands were used for small business.
The question is, was it something like 10,000 or 100,000+ machines.

[luckybuck]

@JamesD

 

Sorry for the late answer, last 2 days were heavy.

 

Thanks for the long answer, yes, you are right. I guess we had the same problem with a good Pascal in those days. Atari tried this here:

 

 

CX8109 Pascal

CX8110 Pascal Database

CX8115 Pascal-Native Code

CX8116 Pascal-Linker

All 4 disks were planed, but not published due to my research, because they needed 128 KB RAM and 2 disk drives. In the beginning(!) of Atari...

Only APX Pascal made it.

Can very good imagine, that this would be the same, when thinking of Fortran... But I am just guessing, not knowing.

[luckybuck]

@ricortes: Yes, IBM was too big in those days, the 'game' computers were not taken serious. But sometimes, I have that feeling that there was an IBM-man at the very top of Atari, to make sure, they stay in the games area...

 

Too many products from Atari were finished, but published.

[luckybuck]

@Larry: well, i would love to check my f77-programs from the 80's whether they run on the Atari. Just for nostalgia reasons and of course fun.

[luckybuck]

@JamesD:

"

Almost everything Atari released was game or home user oriented. It was Apple themselves that ported UCSD Pascal and Fortran to the Apple II. There's no reason Atari couldn't have done the same.

"

Thinking exact the same. Especially for small business. :-)))

 

That is, what I had in mind. :-)))

 

Thanks.

[luckybuck]

@JamesD:

"

Out of that it would not be unreasonable to assume thousands were used for small business.
The question is, was it something like 10,000 or 100,000+ machines.

"

Yes, that is what I would like to know. Aren't there any IT historians? Does this discipline even exists?

[JamesD]

@Larry: well, i would love to check my f77-programs from the 80's whether they run on the Atari. Just for nostalgia reasons and of course fun.

Fortran itself doesn't have any special requirements.

As long as the compiler isn't a subset, the code should run.

You might have to reformat the output for a different screen width but that's about it.

 

[JamesD]

@JamesD:

"

Out of that it would not be unreasonable to assume thousands were used for small business.

The question is, was it something like 10,000 or 100,000+ machines.

"

Yes, that is what I would like to know. Aren't there any IT historians? Does this discipline even exists?

That would have required some foresight. Nobody had any idea sales would take off like they did.

Tandy even thought they could always use their machines for their stores if they didn't sell.

Plus, almost all these companies were small, and privately owned. There were no regular stockholder sales reports.

What info has come out is usually from what people remember.

 

The problem with the sales numbers we know is that a lot of those machines were initially 4K.

I'm sure upgrades rapidly followed but what they were used for would have been limited at first.

I think 8K to 16K and the intro of Microsoft BASIC (1975) was the turning point for CP/M but BASIC was in ROM on the "trinity" machines so that was a lot of RAM at the time.

 

I'm guessing 1975-1976 was when 8 bit computers started to really look useful rather than just as a toy.

People could easily write their own software to do whatever they wanted if they had enough RAM and a little time.

Computer owners exchanged their programs at club meetings and soon a market popped up with small applications written in BASIC.

If you have ever seen the Softside catalog it was filled with a lot of BASIC programs.

Yeah, a lot were games but there were also business, scientific, statistical, etc... programs.

At that point you have people starting to use machines for repetitive tasks in their business but it would probably be more mom & pop stores or isolated cases where someone could slide it through without approval from their boss.

 

Development tools such as editors and assemblers were common on CP/M by 1976.

Wordstar (developed on a Northstar) was released in 1978 brought professional quality word processing to the 8 bit world.

Visicalc came out in 1979.

I think that's the point where people started really pushing computers for business and businesses other than mom & pop places could widely justify the cost.

There were certainly some bleeding edge visionaries with custom applications before 1978 but numbers would have been limited to possibly tens of thousands rather than hundreds of thousands of machines.

Just being able to duplicate the functionality of much more expensive word processing machines of the day would have resulted in a lot of sales after the intro of Wordstar.

 

As for Fortran... UCSD Pascal came out in 1973(?), possibly inspiring Tiny Pascal which was published in 1978.

I can't find an exact date where it was released on CP/M but it would have been around the same time.

I think that's when people realized compilers were viable on 8 bit machines and you didn't have to resort to assembly for speed or unprotected BASIC.

That's when the floodgates opened for compilers. Small C, Fortran, ADA, etc...

Apple Pascal was released in 1979 followed by Apple Fortran in 1980.

 

 

[luckybuck]

Ah yes, you are right, that is the same with merchandising with Star Wars, no one took that serious, but made George independent. :-)

 

Yes, definite, 1979 with Visicalc was a turning point.

 

https://en.wikipedia.org/wiki/UCSD_Pascal

 

1978, uff 5 years later. Good things take their time.

 

Pascal was from 1970:

 

https://en.wikipedia.org/wiki/Pascal_(programming_language)

 

Apple Fortran in 1980, didn't knew that, but there must be a source code then for the 6502... :-)

 

JamesD, then we have all the facts. From 1979 on, the 8 bits were taken serious for small(!) business.

 

As you wrote: Wordstar in 1978

 

Combined, in my opinion, all facts were on the table, so sad Ridley Scott's Atari neon light insert in Blade Runner from 1982 with regard to 2019(?) didn't come true, because of false decisions... :-(

[JamesD]

UCSD may have come out in 1978, but that was the PDP version, not an 8 bit version.
I've spent a lot of time looking at UCSD Pascal and the dates are a bit hazy.
If Fortran is like UCSD Pascal, the compiler itself is P-Code and you would just need a P-Machine and possibly a few extensions to run in on another machine.

[JamesD]

After poking around Flex sites I seem to have stumbled across source to a Fortran compiler.
It appears to be written in Pascal. The source appears to be over 140K in size and the executable appears to be almost 44K.

This version appears to be from 1983. I'm not sure what the original release date was.
The fact that it's in Pascal does seem to indicate that the release of Pascal compilers for 8 bits set off the release of more compilers.

It is a native code compiler not a P-Code compiler btw.

There is also source for Pascal, Cobol and a C compiler. Only the C compiler is written in C.
There is an accounting package which seems to have some parts written in Pascal but most are in C.

Edited July 20, 2015 by JamesD

[luckybuck]

Thank you JamesD for all your help in this case. :-)

[luckybuck]

Source:

 

http://apple2history.org/history/ah17/#03

 

 

FORTRAN

Released by Apple in 1980, Apple FORTRAN ran under the Pascal operating system. It cost $200 (over and above the $495 needed to get the Language System). Programs written in FORTRAN for other computers could run with little modification under Apple FORTRAN (if a user needed that ability). It compiled to a similar code as did Pascal programs, so was not any faster than Pascal. Apple’s version of FORTRAN had many bugs in it, though, and after its introduction in 1980 it was never upgraded. By September 1986 it had disappeared from their product catalogs.

Another way for an Apple II user to get FORTRAN was to buy the Microsoft Z-80 Softcard for $345 and Microsoft FORTRAN for $200. This version of FORTRAN was more full-featured than Apple’s, and offered some advantages in usability. It did notrequire changing to the 16-sector disk controller ROMs (if you didn’t want to). Also, standard Microsoft BASIC (which was more advanced than Applesoft) was included in the Softcard package.^[9]

In June of 1987 Pecan Software released FORTRAN for the IIGS. It ran underProDOS 16 (GS/OS), but still used the UCSD Pascal disk format for its FORTRAN by creating a ProDOS file that then acted as a UCSD Volume.^[10]

[JamesD]

As I said, if you want to port Apple FORTRAN you'll have to port the P-Machine.
I've looked at porting it to another system so I have spent some time on this.

UCSD Pascal is basically a similar idea to CP/M. A common environment, common applications, and a common CPU. But in this case a virtual CPU.

Apple licensed UCSD Pascal and released it as Apple Pascal. It was one of the best ports.
UCSD Pascal was released on the OSI so you can also look at it.
You can find a disk image for OSI here but I couldn't find a description of the disk image format.
http://pascal.hansotten.com/index.php?page=files
It appears to be just raw tracks and sectors but I didn't mess with it further.

There is a disassembly of it the Apple P-Machine out there. The site it was on called it source code but it was far from that state.
I have no idea what site though.
You have to replace the the I/O section (video, disk, etc...) and you'll have to look at any custom extensions that come with Fortran.
Apple Pascal supported native assembly extensions.

If you get the environment ported, Pascal, editor, etc... should all run.

[luckybuck]

Thanks again JamesD for all your excellent help. I really appreciate that. :-)

[JamesD]

If someone wants to port the 6809 compiler you can find it here:
http://www.rtmx.com/UniFLEX/

Just remember the 6809 has smaller code than the 6502 so a 44K compiler running on the 6809 would need parts of the compiler paged in and out to run on the 6502,
Plus the code generator would have to be much larger to support the 6502 because it lacks 16 bit support and a lot of the 6809's other features such as stack relative addressing.

[Wrathchild]

Is the aim to have a native toolchain, or cross compile F77 code to target the 6502?

If the latter, maybe the Open-Watcom sources could be a start?

[luckybuck]

@JamesD: Thank you, every step forward is welcome. :-)

 

@Wrathchild: Well, cool would be of course a SuperCart with maybe a disk optional native on 6502. A Cross Compiler, which produces native 6502 code would be good, too. But the solution from MS regarding Fortran 80 on the Apple II to insert a Z80-card and then(!) run Fortran, is not what I have in mind.

 

Well, at the end of the 80's we had: Pecan Software FORTRAN apple ii, up to now, I couldn't find the images nor a manual. But if someday, we can get them and maybe a source code, then it should work, doesn't it?

[JamesD]

Pecan Software's Fortran was for the IIgs. That's a 65816 which is much more like the 6809.
With 24 bit addressing, stack relative addressing and 16 bit support you could port the 6809 Flex compiler fairly quickly.
But you'd need a 65816 board in your Atari to run the code.

The Wacom Fortran source archive is huge.
I'm guessing a lot of that compiler is geared towards optimizing PC code and would have to be scrapped because the CPUs are totally different.
It would have intermediate code optimization like most modern compilers where the Flex compiler probably doesn't.
It would probably be better to port the GCC Fortran as it was designed to have code generators for different CPUs.
I spent some time with GCC's internals and it would probably take longer to get familiar with than it would to finish the port of the Flex compiler.

[luckybuck]

Uff, didn't knew that! Thank you so much JamesD. Had never an Apple 8 bit, only in school, therefore.

 

So, then Fortran on 6502 is complete new land. :-)

 

That takes time and work. Good, that you could make the facts clear. Thank you, that was a great help. :-)))

[danwinslow]

I want somebody to port Algol!

[JamesD]

I want somebody to port Algol!

What would you compile?

 

Pascal has a similar syntax with some notable exceptions and is more portable.

[luckybuck]

Well, Pascal is a the successor of Algol, isn't it?

 

Anyway, would be glad to add that on the Atariwiki. :-)

Edited July 21, 2015 by luckybuck