SMS2 and STELLA

[RalfR]

Will Stella run Quill?

Possibly. But without a way to execute it, it will certainly be difficult.

[Derek_Stewart]

Will Stella run Quill?

And the prize for most important question goes to Derek!

OK, thanks for a support here...

Maybe I should said, will Stella execute a QDOS job, Quill being a common QDOS executable programme that most people can use.

[Tinyfpga]

So far, the responses to this topic have been useful. To date they have all been negative. Any takers
for a positive response? I get the impression that no one has read Peter Graf's reply to TT's post in
March 2017.

I have never suggested that it would be useful (or even possible) to place Stella within a conventional
computing environment. That would have been ridiculous 25 years ago, today it would be absurd.

Ever since the creation of tinyfpga boards ( there have been quite few designed for hobby use) I have
seen a realistic future for stella.
When Peter posted a picture of his Qzero board I thought:- "finally, and for the first time in 25 years, here
is something real in which to place a genuinely interesting OS.

Talk of how difficult it would be to develop Stella in a modern processor is irrelevant because
,as you all know, the Qzero doesn't contain one.
As Peter mentions in his reply:- once a system is written in HDL, one has created hardware
that is unaffected by the vagaries of the market place. This stability greatly reduces the complexity
of an OS.
Even the variation in QL systems created problems for TT's SMSQ. (remember his letter
to QLers?)

Derek asked:- "Will Stella run Quill". A bit of history is required to answer this question.

After the failure of the QL, the subsequent failure of the Sandy Futura and a proposal
to build a lego-like system that was ignored, TT decided to build SMS2 to run on
Atari STs.
SMS2 was not built as a replacement QDOS but as a system to be used by others.
The interpreter was thought to be irrelevant for such users. An early form of SMS2 was produced
to be installed from floppy. This was probably the version that TT sent to Ralf. SMS2 was then
updated to include improvements in the drivers etc. and delivered as a commercial product in
a PEROM cartridge. It is this system that I used for a while and on which I wrote some SBASIC
programs. This may explain why I have not found a use, as yet, for the SMSQE interpreter.

The SMS2 cartridge was advertised and reviewed in various Atari magazines and Jochen Merz sold it
in Germany. As you can imagine, various Atari users asked:- "does SMS2 run
TOS programs?" The answer was, of course, no.

TT was asked to look into this problem and the result was an entirely new OS with a
superset of system calls that could, theoretically, support applications from any domain
That OS he called Stella.
At the time of its creation Sella was going to support Atari and SMS2 applications.
Support for QL applications was not considered important. There is a version of xchange that
runs on SMS2

I can not imagine a version of Stella in FPGA in which one would want to run Quill.
SMSQE in FPGA takes care of that need.

I have made a list of positives for Stella that currently outnumber the negatives. I will to put them
in a pro-Stella document.

[Peter]

Now there's a challenge. I will give it a try at some point. How about Peter Graf likes it (or maybe he doesn't).

I can't tell if like Stella. I never had an opportunity to try it.

I just see a last niche for a small and efficient 68K assembler based OS (with decent C language interface for applications) in resource constrained FPGA usage. There are FPGA applications where one wants to add a small on-chip microcontroller with embedded operating system. But does not want to develop under the restrictions of an 8 Bitter, nor the code size waste of a smallish 32 Bit RISC CPU softcore. And would appreciate the OS being very small, so everything fits into internal block RAM.

[Peter]

When Peter posted a picture of his Qzero board I thought:- "finally, and for the first time in 25 years, here is something real in which to place a genuinely interesting OS.

Thanks for the flowers. I would admit the Qzero size/cost effectiveness was a bit ahead of comparable tiny FPGA boards at design time. But the Qzero has aged for three years already, and it's potential will probably never be used, because I only work under hobbyist conditions. An OS like Stella would better be based on a different FPGA platform, maybe one without extenal SDRAMs so it can take full advantage of small code size.

[NormanDunbar]

Any takers for a positive response?

I replied, on the other thread I think, that I was interested in Stella and wanted to try it out waaaaay back when I read about it. However, the silence has been deafening since then -- until you mentioned it -- and as Tony has retired to France to live out his days in splendour, it's unlikely that it will ever see any more daylight than you are giving it on your Atari. Other than you, I know no-one who has even seen the actual OS itself, it's a shame as Tony had some bloody good ideas back then -- with "back then" being the operative words.

Tony also had some weird ideas too, OO programming is bad, very bad, so bad it's damned stupid, but I paraphrase, but like everything else, apart from domestic cats and Wordle, there's good in any programming language, well may Brain F*ck is the exception there!

When I say "weird" I mean that I thought them weird. I know some others here hate OO programming with a vengeance, but I have an application out there in the wild, Trace Collier at https://github.com/NormanDunbar/TraceCollier, being used by Oracle database Professionals, which I originally wrote in C but converted to C++ as it was easier to code, and more importantly, to maintain, in OO form. Tony's "blanket ban" on OO is similar to all the Basic purists who complain about the slightest mention of a GOTO, or C programmers who can't stand anyone even thinking about using a global variable!

Cheers,
Norm.

[XorA]

I can not imagine a version of Stella in FPGA in which one would want to run Quill.
SMSQE in FPGA takes care of that need.

This one thing you keep saying that I just cant get my head around, why does an OS need to be encoded into the RTL of an FPGA? Surely that defeats the purpose?

[Derek_Stewart]

Hi

I think you have missed my point, lets make a rash assumption, lets imagine that Stella is made wirk on a hardware platform.

How many applications are available to use on the new Stella Operating System?

Back to the real world, where we have SMSQ/E v3.38 and probably the best programmers in the QL World, to maintain the operating system, that can run all QL applications, either new or old.

Personally I can not see the point of developing Stella, maybe if the source was made public domain, it could stand chance.

[RalfR]

An early form of SMS2 was producedto be installed from floppy. This was probably the version that TT sent to Ralf.

That could be possible. It was an Atari disk, which produces(!!) an SMS2 startup disk. It was not the SMS2 startup disk itself, which he has sent to me.

There is a version of xchange that runs on SMS2

This was just a patched version from TT to work with b/w screens. It also runs (as usual) on QDOS systems. Nothing special.

[Tinyfpga]

When I write:- "Stella or SMS in FPGA", I mean Stella or SMS running in an FPGA implementation of a
"system-on-chip" and not an OS as a digital circuit, although TT did once suggest that such a thing would be very fast.
The Q68 I would describe as SMSQE in FPGA. I did not realise this would be misleading.

I had to look up the meaning of Register-Transfer Level. I only vaguely understand the Wikipedia entry.

As Derek says:- "In the real world we have SMSQE". To this I reply:- "and I am very pleased that we do."
He then creates a tiny chink of light with his "it could stand a chance".

Curiously, I have somehow created an SMSQE environment where almost no QL programs will
run properly or run at all. Not that I mind because I use SMSQE like I use SMS2, and that is, to
write SBASIC programs and nothing else. The experience so gained, allows me to write programs
in Visual Basic and PYTHON. None of this is actually useful, of course.

I was invigorated by both pjw and norm to try and use the SMSQE interpreter as described in their posts.
The result was always a scrambled screen. I was unable to duplicate their experience.

I am not going to count all the application programs I have on my 8 SMSQE "hard discs" but there are
about 200. The current ratio of my own programs to "QL" programs is about 10:1

As mentioned in my previous post I am putting together a pro Stella argument. I imagine a novel computing
environment where one writes one's own programs.