_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

There does seem to have been a bit of culture change on this forum in the last month or two. Both regulars and newcomers seem to be happy to make challenging statements of opinion as if they are spoiling for an argument. That's not usually how we proceed here, although we've had our share of disagreement.

It helps, I think, to qualify claims as being your own thoughts or opinions. (When I say "you" or "yours" I mean anyone here.)

It doesn't help to take a position that some technology or product is dead, or not viable, or uninteresting. Any one of us might have such an opinion, but stating that kind of opinion doesn't allow for any kind of illuminating or productive discussion.

Especially claims that something is uninteresting, or not useful, or incomprehensible - those are negative claims which simply don't belong. The forum is not here for such a narrow purpose that all of us can find every subject interesting. If you don't like a topic, just don't follow that topic. There are plenty of others. If you'd like to understand something, but don't yet understand it, by all means try to gain understanding through discussion.

Well, just be aware that Forth Inc. will do everything in their power to discredit the 65VM02 or any other Forth that anybody comes out with. The whole purpose of Forth Inc. writing the ANS-Forth standard in 1994 was so they could say that the competition is just a non-standard wanna-bee. It is not a coincidence that my MFX was written in 1994/1995 and that I wrote MFX in UR/Forth, and that MFX and UR/Forth were both discredited by ANS-Forth as being non-standard and instantly obsolete. Also, be aware that Forth Inc. tried to get that job at Testra writing the development system for the MiniForth processor. Testra hired me instead, and Forth Inc. has never forgiven me for stealing the bread and butter from Forth Inc.'s table. Elizabeth Rather said (on the Forth-200x mailing-list) that I got fired from Testra because the MFX was no good. That is a lie --- Testra continues to use MFX today (their RACE is just the MiniForth upgraded from the Lattice isp1048 PLD to an FPGA).

Anyway, I have a new update on the 65VM02 design. I got rid of the VIRQ which was over-complicated, and I now have a simpler and better design. I also figured out a way to significantly speed up ENTER and LEAVE which should speed up all code that uses local variables. What I have now is, I think, as efficient as is reasonably possible --- there is no further opportunity for new instructions that do things in parallel internally.

I have no intention of supporting JVM --- Java isn't my cup of tea --- also, Java is 32-bit, so I think it needs a 32-bit processor.

The 65VM02 is for applications that are bigger than the typical 8-bit processor program with 128 bytes of memory --- but are not as big as what you would use the STM8 for --- not too small, and not too big.

Note that I previously described 8-bit processor programs with 128 bytes of memory as being "uninteresting" --- I wasn't being disparaging, as this is a worthwhile business (most micro-controllers are in this category and that is where most of the money is) --- I just meant that it is uninteresting to me because it has nothing to do with the 65VM02 design.

Yes, I know that.

The 65VM02 is fully 65c02 compatible and should run all legacy code just fine. INC and DEC are the same as before, as are all the 65c02 instructions. I even left in the weird bugs involving the return-address being one off, and the S register being one off --- there is likely legacy code that works around these "features" and would be broken if this was changed.

I just commented that INC and DEC failing to affect the carry-flag seems like a flaw to me. Certainly I had to work-around this in the CMOVE function, which makes the code bloaty and slow. This was my comment in the code:

Well, don't be over-impressed by a website. The internet is full of websites that appear to represent "very professional" companies that are in a "higher league" than everybody else.

Only once has anybody shown up on comp.lang.forth claiming to have written a commercial program in SwiftForth. This was Roger Levy and this was the program:
http://store.steampowered.com/app/341060/The_Lady/
If this is a higher league, I think I'll stick to the lower leagues...

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I added these to the 65VM02:

So, I've got Y covered, while still leaving DEY and INY as they were. I don't have anything like DEC and INC that affects the C-flag though, which is why I had to work around that problem in the CMOVE primitive. I suppose I could provide new instructions there too, but I don't want to add too many new instructions, and this is a minor issue.

As for DEX and INX I just use X for the data-stack pointer so it never rolls over, so these are fine as they are.

I don't really know very much about multi-tasking OS design. I've never used one --- it is just something I've read about, and not very much.

Previously I had the T flag and I assumed that there could not be a task-switch in the middle of a primitive, but a task-switch could only occur in NEXT which is between primitives.

Then I got rid of the T flag. My thought was that only ISRs would access I/O ports, and they would send or receive the data to or from the tasks in circular buffers. It is okay for an interrupt to occur in the middle of a primitive even if the primitive is accessing a circular buffer and the ISR accesses the same circular buffer. So, no worries!

It may be necessary to reintroduce the T flag though, and have it set by the heartbeat timer and checked in NEXT, and make the rule that there can be no task-switch in the middle of a primitive, but only in NEXT. This way a primitive would effectively exclude any other task from accessing the same datum that it is accessing.

Can anybody recommend anything that I could read on the subject of multi-tasking OS design? This is a subject that I need to learn more about before I get further into the 65VM02 design.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Maurice Bach's Design of the UNIX Operating System is the classic tome on the subject. I purchased my copy in 1987 and have read it cover-to-cover several times. The book does not focus on any one machine architecture, which means you'll get a more balance view of the theory behind a proven preemptive, multitasking design.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

I'd go for Tanenbaum! Modern Operating Systems on the one hand, and Operating Systems, Design and Implementation on the other.

At one time I had this instruction:

The purpose of this was to read and write a semaphore in a single instruction so it couldn't be interrupted in the middle (this works with an absolute address rather than a direct-page address because the direct-page is task-specific).

I may go back to EXA so I can have a primitive that reads and writes a semaphore and can't be interrupted. This would allow code such as you are describing (accessing an I/O port from a task and being sure that you don't get a task-switch in the middle of accessing the port) --- and this code could be written in Forth.

I still don't like the idea of a task-switch being only on NEXT because this slows down NEXT --- also, this means that accessing an I/O port has to be done inside of a primitive, which means that it has to be written in assembly-language --- it is much more convenient if this could be written in Forth using C@ C! etc., but this involves multiple primitives.

Thanks for the links to information on multi-tasking OS design, from you and the others --- I will read up on that.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

You could make different interrupt priorities, where the highest priority would be handled in assembly language, and can happen at any time, and a lower priority one that works on Forth level, and can only happen during NEXT.