Hi there, the next stage of my little Orwell computer is getting BASIC running on it. After reading the Pagetable blog post on it I decided to go with Microsoft BASIC. It is also in keeping with my whole 80s computer theme.

In addition to the code there I found Grant Searle's version of it he used on his simple 6502 machine. His code is all the relevant bits of the OSI version of BASIC in one file with macros and unnecessary ifdefs removed. I took this code as my start point.

I added in my own IO routines and after a few minor bugs got it working. I can get past the Cold/Warm restart page but then it all goes wonky at the Memory Size prompt. The symptom is odd behaviour in the key values I get returned. For example I can return an 'a' but 'A' comes through as garbage. No matter what I enter here I get a syntax error and the prompt repeats. I think somehow something is getting corrupted. In order to understand what I have been trying to understand the code but something is confusing me.

I believe things are going wrong in the COLD_START code.

That cmp #$41 is actually a kind of easter egg. An 'A' typed here is supposed to show who wrote the code. But I get garbage not an 'A' at this point.

I have been trying to understand how the jsr CHRGET works. But I don't see how CHRGET is a subroutine? In code it is declared like this:



That's actually in the zeropage (at $BC).

Later in the actual code there is this:


I don't understand how this works? How doe we get into that code? I guess TXTPTR is a buffer for the value entered from the keyboard but how does all this get called exactly?

All my code is too big to post and I don't have it online yet but Grant's version is here.

My next thing to try is take all my code out and just hard code things just so I can see if I can get past the memory size/check stage with the standard OSI code. I don't know if any of the above is part of my problem but it would be nice to understand how it works!

Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer

What's going on is actually kind of cute... Or perhaps cutesy. GENERIC_CHRGET through GENERIC_CHRGET_END is copied to the zpage at some point during startup (the code wouldn't work, otherwise). The definitions for CHRGET, TXTPTR, CHRGOT, CHRGOT2, and RNDSEED are all defined as zero-page locations, all relative to CHRGET (the + CHRGET term at the end of most of the definitions) based on the displacement of the GENERIC_ version of the symbol relative to GENERIC_CHRGET. This allows for a default value to be seeded for the random number seed, and the CHRGET routine itself gets to use LDA absolute to read characters, with the absolute address in the instruction being stored at TXTPTR, which is a speed optimization.

Now that I've seen this trick I'm going to have to think carefully about if I can (or want) to use it in my own software.

Ah, the moving is done in the code just above that I posted above. In the cold start routine which makes perfect sense of course. Has to be done once!



Makes more sense now (although I still don't 100% follow). I still have issues with my code. I did try moving my variables to another part of memory well away from the ZP stuff but still the same issue. I did verify my actual key scan is returning the correct key though. Will keep digging!

Thanks!

Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer

Well, I finally worked out my bug. The line reading code uses X to keep track of where it is up to on the input buffer. There is a check to make sure that X hasn't fallen off the end of the line in which case the last character isn't added to the buffer. Unfortunately my keyboard scanning code also uses X. I use it as an offset into my key matrix look up table. The matrix isn't in any sensible order from a character point of view.

What was happening was certain characters were causing me to use an X offset into my lookup table that changed X enough that the character wasn't being stored in the line buffer. That's why some characters worked fine but not others with no obvious pattern.

Fixed by preserving X before I use it of course then restoring it later.

And that worked! I can now boot up and get into BASIC. It runs the memory check and so on. I can enter lines but so far can't run anything. It just says 'OK'. Typing 'LIST' returns nothing but 'OK'. Next issue to work out.

The other thing I need to do is add in support for my alpha lock key as using shift to type everything in upper case is annoying!



Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer

Or you could change the keyboard lookup table.

Well, the lookup table directly maps to how the matrix is laid out to make decoding it in software easy. I scan the rows and read the columns in one byte at a time so it makes the lookup easy. I can just say something like key position = (row * 8) + column to find the key. Now I save X before mucking with it it's all fine. It just took me a bit of digging in the OSI code to realise they were using it too. Debugging like this is a great way to learn what their code is doing of course.

I now have alpha working and commands are working (I can poke into my debug buffer and light up LEDs and "? FRE(0)" returns correctly) but remembering of line numbers isn't working for some reason. It's not storing my lines of code. No errors though.

Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer

Well, finally I can report success! I now (after weeks of debugging) have Microsoft BASIC running on my Orwell machine.

Many times I was going to post here for help but in the end I kept plugging away. I wrote routines to dump out what was in memory. I compared my code to the original OSI code, original Applesoft code, whatever other versions of the code I could find. I even downloaded an OSI machine emulator so I could see what it's memory looked like. I have a much better (but not complete) understanding of how OSI BASIC works now.

In the end my bug was simple. The code to check for control-c seems broken. It calls into ISCNTC routine then no matter what it just runs into the stop code. So the effect on any LIST or RUN command was as if control-C was hit right away. Interestingly the code on PageTable seems to note this in the comments in the osi_iscntc.s file (wish I'd seen that weeks ago!). I started my code from the version Grant Searle has on his site. His code seems to just run into stop too. I changed mine so if control-c isn't hit it just returns and keeps going. Maybe I am missing something subtle here but it seems to be working for me now.

A few things that really helped were http://www.osiweb.org/manuals/basicnotes.pdf and http://www.osiweb.org/misc/OSI%20ROM%20Basic%20Datasheet%20from%20Aardvark%201979.pdf. Those notes really explain how some things are done (like the line parsing and program storing in memory).

Still a lot to do. I want to keep studying the code and, more importantly, comment it. I'll never remember how it works otherwise. I need to come up with some load/save routines. And I need to keep working on my Eagle layout for a real circuit board for the thing.

But I can now start reviewing all my 80s Usborne books and actually use the machine to enter in all the program listings now.

This has been bugging me for so long now it's really a relief to finally get it working. I've literally been losing sleep over it.

Here are some pictures:





Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer

Very nice! Make a note in your copy book.

Ah, I see what was causing your problem. Not enough red wire on the breadboard!

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

Ha! I could fix that! But now I need to sort out a real PCB for it I think. And I might add a power on reset circuit.

I spent ages last night after I got it running just writing small BASIC programs poking memory to make the piezo make silly noises. I am using a VIA to do my keyboard matrix scanning and since PB7 was unused for that I figured I might as well use it for something!

Is it enough to put 0 into the timer one latch to stop it or do you actually need to disable it?

Funny thing is I remember the first thing I did with my Vic 20 (first machine way back when) was the little programs in the manual to make noises.

Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer

Cool. Ehat kind of keyboard you are using? It looks really small but with "normal" sized keys. I have recycled an old C64 keyboard, but on yours there are the mostly needed characters. The C64 keyboard has only a lot of specific graphic symbols that do not need.

Mario.

_________________
How should I know what I think, until I hear what I've said.

Looks like http://www.oldcomputermuseum.com/ti_994a.html

Exactly right! I got it off eBay a while ago. Apparently RadioShack sold them years ago as a surplus part. Mine came with the original packaging! It's a very nice little keyboard actually. Proper keys! Lovely to type on but compact.

Simon

_________________
My 6502 related blog: http://www.asciimation.co.nz/bb/category/6502-computer