| Author |
Message |
|
|
 Posted: Sep 22, 2008 - 04:08 AM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
I have created a Basic interpreter for the AVR mega devices and would like to post the .hex file on the AVR Freaks site.
My goal with this project is to make the AVR devices accessible to those not skilled in C but wanting to use the chips in their projects.
KLBasic is written in C and lives in the 'mega's flash. Your Basic program is stored in RAM as tokens. Your Basic program can be saved to EEPROM or flash.
Note that this is a 32-bit integer Basic.
On a 90can128 at 20 MHz, this Basic executes 50,000 empty loops (FOR-NEXT) in 1.088 seconds; not too shabby for an interpreter.
Because the 90can128 and similar devices only have limited RAM, you will not be able to write really large programs. However, if you can get the atmega1284p, with 16K of RAM, you should be able to write substantial programs.
Features of KLBasic:
32-bit integer,
support for single-dimension arrays,
variable names significant to 8 characters,
SAVE/LOAD to/from EEPROM or three on-chip flash files,
autostart (runs any stored program on power-up),
direct access to I/O ports (no PEEK/POKE needed),
small footprint (the 90can128 version uses 29KB of flash)
This is just a quick overview of KLBasic. I'm working on a user's manual but for now you'll just have to play with it. I would appreciate any feedback.
Note that the .hex file is for the 90can128 but I believe works also on the atmega128. It does NOT work on the atmega1284p yet; I need to rebuild and test for that device.
BTW, this is my first post. If something gets lost, I'll try again. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 04:13 AM |
|
Joined: Dec 11, 2007
Posts: 6982
Location: Cleveland, OH
|
|
Klaxon44,
Welcome to the Forum!
Competition for the Basic Stamp, how fine!
JC |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 04:25 AM |
|
Joined: Jun 04, 2007
Posts: 1075
Location: Cincinnati, Ohio - USA
|
|
| I'm afraid I've never seen the appeal for an interpreted basic on a micro when basic compilers are available. It seems you would need to download the program either way. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 04:33 AM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
It's all in the immediacy. The interpreter lets you work on your target hardware and enter commands like:
porta = $ff
to turn on the LEDs directly. If it works the way you like, just add:
950 porta = $ff
to your program and you're good to go.
I pay my rent writing in C; have done for years. But there are lots of newbies out there with wonderful ideas but no experience in C and no time/inclination to climb the learning curve. This tool gives them a shot at seeing their projects work. Plus, it's kinda cool to have the entire dev platform for your project IN your project...
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 05:39 AM |
|
Joined: Aug 13, 2006
Posts: 6764
Location: Bellingham, WA - USA
|
|
I for one applaud your work and your willingness to share it, at least in its early stages. I agree that the interactive part of an interpreter is its best feature.
What are your plans? What other devices, what sort of marketing, etc.? Or is this just a work of love by a hobbyist?
Bothel, huh - Microsoft employee? I'll honk next time I drive by on I5.
Thanks for the posting and offer, and welcome to the forum. |
_________________
Chuck Baird
"It's better to catch the trapeze than test the safety net" -- RPi book
http://www.cbaird.org
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 05:54 AM |
|
Joined: Mar 28, 2001
Posts: 20635
Location: Sydney, Australia (Gum trees, Koalas and Kangaroos, No Edelweiss)
|
|
Hurray!! No more C vs ASM wars...  |
_________________
John Samperi
Ampertronics Pty. Ltd.
www.ampertronics.com.au
* Electronic Design * Custom Products * Contract Assembly
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 05:54 AM |
|
Joined: Jun 04, 2007
Posts: 1075
Location: Cincinnati, Ohio - USA
|
|
|
Chuck wrote:
I agree that the interactive part of an interpreter is its best feature.
Maybe I missed something somewhere. It seems like when I played with a Basic Stamp years ago you had to compile it to pcode and then download it to the stamp.
I don't mean to sound as though I'm hammering his efforts, just trying to understand better. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 06:15 AM |
|
Joined: Aug 13, 2006
Posts: 6764
Location: Bellingham, WA - USA
|
|
|
Quote:
Hurray!! No more C vs ASM wars...
Quote:
"No, not a tape measure, you nitwit! A screwdriver!"
"All that's here's a hammer. Will that do?"
The samperi looked at his well meaning but inept assistant in disdain. Good help is so hard to find, especially when you're too cheap to pay them.
He could read no more, he felt such an identification with the hero. His eyes welled up with tears as he closed the book, tidied himself, and absently wandered back to the major throne room from the minor throne room.
The poor hero. "What could be worse?" he thought. Then it struck him. No true index register. |
_________________
Chuck Baird
"It's better to catch the trapeze than test the safety net" -- RPi book
http://www.cbaird.org
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 06:58 AM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
|
klaxon44 wrote:
Note that this is a 32-bit integer Basic.
On a 90can128 at 20 MHz,...
On a mega128, using that hex file you uploaded, is the code set for a 20MHz crystal? I tried it on a 16MHz mega128 and nothing came out UART port 1 at several baud rates I tried. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 09:19 AM |
|
Joined: Dec 20, 2002
Posts: 7279
Location: Dresden, Germany
|
|
Funny to see, even though I'm not one of the potential users.
I once wrote a BASIC interpreter as a proof of concept for my flex
and byacc ports to the AVR. It's far from being perfect in any sense,
and obviously makes most sense when running on an AVR with external
SRAM to store the program lines into. I eventually pushed it to
floating-point numbers and string variables / string concatenation.
No AVR-specific stuff, however.
lex and yacc or not quite optimal for hacking a BASIC interpreter
though, and the resulting code size is fairly clumsy. |
_________________
Jörg Wunsch
Please don't send me PMs, use email if you want to approach me personally.
Please read the `General information...' article before.
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 09:28 AM |
|
Joined: Dec 12, 2002
Posts: 380
Location: Udon Thani, Thailand
|
|
While not a serious replacement for C or Basic compilers (remember Forth?), a basic interpreter is an excellent tool for learning programming concepts, and for debugging hardware -the peeks and pokes and input and output commands are excellent ways to tickle port pins or look at them.
Klaxon44, will you eventually be adding user documentation, and making versions for smaller chips, such as those that come in DIP packages, such as the ATMEGA32 and smaller? |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 10:52 AM |
|
Joined: Jul 18, 2005
Posts: 62944
Location: (using avr-gcc in) Finchingfield, Essex, England
|
|
klaxon44,
Would you not at least consider adding some auto-baudrate detection to it? 20MHz only may severely limit your "market".
(it'd be nice if there were also variants for some of the other AVRs though clearly it's going to need to be a fairly big one)
Cliff
BTW that code seems to include both the app and a bootloader witth the app occupying 29,017 bytes from 0x0000 to 0x7159 then the boot loader occupying 0x1fc00..0x1fcbf. Is the bootloader really required and does that imply that the AVR needs the BOOTRST fuse enabled?. If the interpreter is really less than 32K it'd be nice to get a build for a 32K device. |
_________________
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 03:54 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
In response to several of the above posts:
1. The released .hex file is for the 90can128 but should work on the mega128 as well. Hook to USART0 (PE0 & PE1) at 38.4K, 8N1; your hardware needs a 16 MHz crystal. (My numbers in the original post probably implied 50K loops in 1.088 secs on a 20MHz AVR; it should have been on a 16MHz AVR. Sorry about that.)
2. I definetly want to add autobaud but haven't come up with a clean way to do it. I don't want to use the "wait for user to press CR" style; I would rather have the MCU derive the baudrate values for 38.4K from internal information. Any suggestions?
3. The code in the bootloader area is NOT a bootloader. It is code that allows the Basic interpreter to write Basic programs into lower flash pages. For example, you can save a program into lower flash with "save fl0." This will write your Basic program into flash file 0; it can be recovered with "load fl0" or autorun with "autost fl0."
4. I will be adding various chips in the coming days/weeks, and will provide a suitable user's manual. The core of this program is target-independent; creating a KLBasic for any device is limited to writing a single device-specific file that accesses various types of memory, handles loads/saves, and handles I/O ports. In fact, there is an equivalent KLBasic for Windows (console app) that I use as a development tool. I get the Windows version running first, then port the (minimal) changes to the AVR file and I'm done.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 03:58 PM |
|
Joined: Jul 18, 2005
Posts: 62944
Location: (using avr-gcc in) Finchingfield, Essex, England
|
|
|
Quote:
I don't want to use the "wait for user to press CR" style;
Why not? - that's the way it's usually done. In fact modems are only able to auto-baud because they expect to see "AT". but having the user hit [ENTER] ensures the start-bit is distinct so can be measured - but it helps to route this to an INT or ICP pin which maybe slight complicated the wiring (and wastes a pin) |
_________________
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 04:01 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
To address CirMicro's post...
There are some interpreters that compile off-target to pcode, then require you to move the pcode into the target. This is not one of those interpreters.
In KLBasic, you work solely on the target through a comm program, such as Hyperterm. You enter your program on the target MCU, where it is incrementally reduced to pcode and stored in RAM.
Since the translation is on-the-fly, you can pretty much type in any Basic statement without a preceding line number and the statement will be interpreted and executed immediately.
This lets you do things like modify the I/O ports at the command prompt, just to see what will happen.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 04:06 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
|
clawson wrote:
Quote:
I don't want to use the "wait for user to press CR" style;
Why not? - that's the way it's usually done. In fact modems are only able to auto-baud because they expect to see "AT". but having the user hit [ENTER] ensures the start-bit is distinct so can be measured - but it helps to route this to an INT or ICP pin which maybe slight complicated the wiring (and wastes a pin)
I want to avoid any user interaction at start up because KLBasic can be used to autostart an embedded app, such as a robot or greenhouse controller. Such a startup may require immediate transmission on USART0 with no ability of the user to press a key for autobauding.
I will (very reluctantly) use a variation of this autobauding if that is simply the only way to do it. But I would MUCH prefer to have the MCU compute the baud values on-the-fly without user interaction.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 04:19 PM |
|
Joined: Jul 18, 2005
Posts: 62944
Location: (using avr-gcc in) Finchingfield, Essex, England
|
|
So have it so that either:
a) it checks for a link via a PIN
b) it looks for a trigger value in an EEPROM location
that makes it stop and wait for the user interaction to auto-baud. The user like me will then program the EEPROM or wire a link to Vcc or whatever to say "don't just run, stop and wait for <CR> to auto baud rate"
(in fact links are best done by enabling the pull-up on the pin to be checked and then looking for it being connected to Gnd.)
Cliff |
_________________
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 05:00 PM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
|
klaxon44 wrote:
In response to several of the above posts:
1. The released .hex file is for the 90can128 but should work on the mega128 as well. Hook to USART0 (PE0 & PE1) at 38.4K, 8N1; your hardware needs a 16 MHz crystal. (My numbers in the original post probably implied 50K loops in 1.088 secs on a 20MHz AVR; it should have been on a 16MHz AVR. Sorry about that.)
FYI: When I ran the code on the 20MHz mega128, nothing at all came out UART0. I would have at least expected some garbled characters (speed mismatch). After running that, I immediately flashed some code I have for 16MHz and UART0 is working fine. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 07:19 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
|
stevech wrote:
klaxon44 wrote:
In response to several of the above posts:
1. The released .hex file is for the 90can128 but should work on the mega128 as well. Hook to USART0 (PE0 & PE1) at 38.4K, 8N1; your hardware needs a 16 MHz crystal. (My numbers in the original post probably implied 50K loops in 1.088 secs on a 20MHz AVR; it should have been on a 16MHz AVR. Sorry about that.)
FYI: When I ran the code on the 20MHz mega128, nothing at all came out UART0. I would have at least expected some garbled characters (speed mismatch). After running that, I immediately flashed some code I have for 16MHz and UART0 is working fine.
Darn it! I don't have a 'mega128 to test against. I do most of my work with the 90can128 and the two technical summaries look almost identical so I assumed the 90can128 .hex would just drop in.
I'll see if I can find a 'mega128 dev board somewhere. I'm also going to ramp up my work on the 'mega1284p version. That's really where I want to be, because of the increased RAM.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 09:28 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
I have a problem with the download !
The link is to an index.php file ?!
This make me think back to the microsoft basic to the commodore PET. And they had the full thing into 14K ROM (6502 code). And the very same good part was that you do the calculations directly as well.
At one time I was playing with a AVR emulator running on a AVR (just for fun ) and was able to run the code from an extern eeprom at about 1/10-1/20 of the real speed, so if you need some "speed" input I'm here
We can even meet (I live less than 2 miles from where 405 ends), then Chuck can come as well.
Jens
Edit my location is Edmonds, I don't know why it not showing it. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 22, 2008 - 09:52 PM |
|
Joined: Mar 27, 2002
Posts: 18757
Location: Lund, Sweden
|
|
|
Quote:
I have a problem with the download !
The link is to an index.php file ?!
Thats a known bug in the forum software. The downloaded file is correct, but the suggested name "index.php" is wrong. Just choose to save the file under the correct name and all will be fine. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 23, 2008 - 04:37 AM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
I recompiled for a 'mega128 and have attached that file. The executable expects a 16MHz crystal and talks through USART0 at 38.4K, 8N1. I tested this executable on a BDMicros' Mavric IIB dev board.
I have also attached a short benchmark program so you can get an idea of some of the features of this program. Just do an ASCII text transfer from your comm program; you may need to set a small delay (say, 20 ms) at the end of each line.
Feedback welcome...
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 23, 2008 - 07:50 PM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
|
klaxon44 wrote:
I recompiled for a 'mega128 and have attached that file.
Feedback welcome...
klaxon44
have you had time to document the language? Syntax? etc? |
Last edited by stevech on Sep 23, 2008 - 07:54 PM; edited 1 time in total
|
| |
|
|
|
|
|
|
Posted: Sep 23, 2008 - 07:52 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
First thank to Johan about the name problem.
I don't have a mega128, so can I get the numberes at 16MHz, to make some compare of how I would do it
I see that you don't DIM your var. so do you place them like the microsoft basic (just after the program)
or some kind of stack ?
The same for the FOR NEXT do you have a stack ?
And do you semi compile on the fly (so a goto will be a jump, the microsoft basic compare with current line number and then either start here or at the beginning, so a lot of lines gets slow).
And I don't see real strings, they are a real pain to deal with, especially the gabage collection. (on a commodore64, the clean up can take more than 30min if you make a lot of strings 1 char long, then change them to 2char long then 3 long, now we are running out of memory and stall for 30min)(the way to get round is to force it to clean up more often, (for the MS the free() function will do that).
Have fun
Jens |
|
|
| |
|
|
|
|
|
|
Posted: Sep 23, 2008 - 08:24 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
stevech and sparrow2,
1. No, I haven't yet had a chance to document the syntax or develop a manual. It's on my list because I know how important it is, but I don't have anything yet.
2. Simple variables do not (yet) need to be DIMed, though I will probably add an EXPLICIT command so you can decide whether you must DIM them before use. Arrays must be DIMed before use. All variables are kept in a separate pool of RAM. Use the FREE command to see how much RAM you have available and where it has been assigned.
3. GOTOs and GOSUBs are compiled to pcode + target line number. I'm not happy with this, because the line number search takes place repeatedly during execution. I will fix this and some other foundational issues as time permits.
4. String variables are not supported, though obviously printed constant strings are supported. Additionally, the INPUT statement accepts a quoted string for use as a prompt.
Do these questions mean that some of you have been able to get the .hex file to run?
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 24, 2008 - 10:58 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
I can't test it because I don't have a 1280.
And that will be a problem for your BASIC, it needs to be able to fit in a mega8 or so.
I don't see it as a problem with only 1K of RAM. I remember the ZX80 with 1K RAM and 4K ROM.
And if no need for speed the code can come from a EEprom(like a BASIC stamp).(that was why I started with the AVR emulator so a tiny AVR could run a big program, the emulater take only about 2K flash(depending of how fast it needs to be))
If you want I can help with a ASM version, but as a C freak I asume that isn't want you want
Jens |
|
|
| |
|
|
|
|
|
|
Posted: Sep 24, 2008 - 11:09 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
If you need something for demo where the PC allways is connected you could make the basic run on the PC, and just have a fast RS232(or USB) to the AVR. The AVR then only need a robot program, evt with a automated timer interrupt etc. for the real time events.
Jens |
|
|
| |
|
|
|
|
|
|
Posted: Sep 25, 2008 - 07:54 AM |
|
Joined: Aug 24, 2006
Posts: 89
|
|
Great piece of software! I loaded the mega128 hex file on a 16 Mhz ET-AVR stamp board (from Futurlec) and it runs flawlessly (once I had the usual serial port issues resolved).
Downloading text files using Brey's terminal or Hyperterminal doesn't work very well though (buffer overrun probably), unless someone can point out a "delay after each line" option. So I copied the test program using several "paste to host" operations. Works great, here are the results:
Empty loops per second, uptime comparison: 4474
Empty loops per second, timer0: 5130
Time to execute 50,000 empty loops: 1088
Time to execute 50,000 additions: 5603
Time to execute 50,000 multiplies: 5769
Time to execute 50,000 divisions: 7482
Is there any way to edit lines (as opposed to retyping)?
I see that the original Basic11 has keywords like PORTA, PORTB, APOKE etc. Could you summarize the avr/machine level interfaces?
I live in Kirkland WA myself. With sparrow2, klaxon4 and jrseattle all close by, we could start a local AVR freaks chapter Of course zbaird is invited too.
jrseattle
www.oscilloscopeclock.com |
|
|
| |
|
|
|
|
|
|
Posted: Sep 25, 2008 - 08:50 AM |
|
Joined: Mar 28, 2001
Posts: 20635
Location: Sydney, Australia (Gum trees, Koalas and Kangaroos, No Edelweiss)
|
|
How exciting!! Got it running on one of may boards fitted with the M128.
http://www.members.optusnet.com.au/~amp ... 20diag.gif
I only wish I had some time to play with BASIC, it would bring back lots of memories, at least 4K worth 
I have the Basic11 manual and the .arc file for version 1.55. Had to install arc with winzip to extract it, how quaint!!
I should also have the whole source file somewhere as a have a js version of the .s19 file.
Now what do I do with it? |
_________________
John Samperi
Ampertronics Pty. Ltd.
www.ampertronics.com.au
* Electronic Design * Custom Products * Contract Assembly
|
| |
|
|
|
|
|
|
Posted: Sep 25, 2008 - 04:25 PM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
10 X = 1234
20 GOSUB 1400
déjà vu ! ! |
|
|
| |
|
|
|
|
|
|
Posted: Sep 25, 2008 - 06:02 PM |
|
Joined: Aug 24, 2006
Posts: 89
|
|
Found one potential problem. I used the "autost fl0" command to automatically start a program on reboot, which worked great. Then I turned the unit off overnight and the autostart did not work this morning. The program was still saved in fl0 so I did another "autost fl0" to get it to autostart again.
Unclear why the previous autostart was lost.
jrseattle
www.oscilloscopeclock.com |
|
|
| |
|
|
|
|
|
|
Posted: Sep 25, 2008 - 10:19 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
So a short estimate give about 4000 clk for a empty loop.
semi compiled (where consts is handled as var)
will give something like this
10 x=x+1
could be handled as
stack addr x
stack addr x
stack addr 1
+
=
this should take about 100 clk with subrutine calls.
(the call/ret will be about 50% of the clk's)
20 if x<>1000 then 10 "in MS both then and goto would work"
would take about 50 clk.
so an empty loop would take 150 clk or be about 25 times faster if it's written in ASM. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 26, 2008 - 04:29 AM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
I flashed to a mega128 ETT stamp, 16MHz. UART0 is at 38.4Kbaud
Using Bray's Terminal, I could not upload the benchmark test program. I slowed down the upload to like 5 chars per second. I think there are line feeds in the file and the Interpreter does not want them. I could't find a way to prevent line feeds using the editors I have (several). The interpreter should/could ignore them.
Clue: If I change my terminal program so that when I type CR it sends CRLF, I get the same error when typing manually. That error is ERROR #7: Illegal variable
So I typed in part of the benchmark program and it ran fine.
here's another ditty
100 n = uptime/1000
110 if uptime/1000 = n then 110
120 ?uptime
130 goto 100 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 26, 2008 - 05:26 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
1. There is no way to edit a line short of retyping. I have given thought to adding an ANSI terminal interface, which would allow cool things like source-level debugging, in-line editing, and suchlike. I don't think the overhead would be very much, either; mabye a few K.
2. Sounds like sending a trailing LF might be an issue; I will double-check to make sure that gets stripped out.
3. sparrow2, I agree that if this compiled down to assembly language, it would run about 10x faster and 10x smaller. I purposely chose to do a pcode interpreter rather than a full compiler. One of my reasons was I didn't want to create yet another machine-specific compiler. I may one day do a true target-based Basic compiler, but for now I'm having fun with the interpreter.
4. For the atmega128, you have direct access to all GPIO ports and registers, all A/D-based registers, and all SPI registers. For example, you can write a byte to the SPI with: 130 spdr = $44
and you can read a byte from the SPI with: 140 mydata = spdr
Note that I used the same names as WinAVR uses in their ioxxx.h files. I do not provide names for accessing bits, though.
5. When I use Hyperterm to send my source text into th e target, I include a 5 or 10 msec delay after each line but no delay between characters; works fine for me.
6. jrseattle, not sure why you would lose the autostart from flash following powercycle. Next time you powercycle, enter: autost
at the prompt and see if it says fl0 was marked for autostart. If it wasn't, something is corrupting your EEPROM on power-down or power-up. Let me know what you find out.
7. js, now that you have it running, you can do anything you want with it!
Actually, I intend to provide support for PWM and servos soon, so robotics is right around the corner.
8. jrseattle, sparrow2, and others,
Yes, it would be fun to have a Puget Sound AVR gathering. If you guys want to hook up for pizza and geek-talk some evening, let me know.
So many features to add, so little time...
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 26, 2008 - 05:29 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
1. There is no way to edit a line short of retyping. I have given thought to adding an ANSI terminal interface, which would allow cool things like source-level debugging, in-line editing, and suchlike. I don't think the overhead would be very much, either; mabye a few K.
2. Sounds like sending a trailing LF might be an issue; I will double-check to make sure that gets stripped out.
3. sparrow2, I agree that if this compiled down to assembly language, it would run about 10x faster and 10x smaller. I purposely chose to do a pcode interpreter rather than a full compiler. One of my reasons was I didn't want to create yet another machine-specific compiler. I may one day do a true target-based Basic compiler, but for now I'm having fun with the interpreter.
4. For the atmega128, you have direct access to all GPIO ports and registers, all A/D-based registers, and all SPI registers. For example, you can write a byte to the SPI with: 130 spdr = $44
and you can read a byte from the SPI with: 140 mydata = spdr
Note that I used the same names as WinAVR uses in their ioxxx.h files. I do not provide names for accessing bits, though.
5. When I use Hyperterm to send my source text into th e target, I include a 5 or 10 msec delay after each line but no delay between characters; works fine for me.
6. jrseattle, not sure why you would lose the autostart from flash following powercycle. Next time you powercycle, enter: autost
at the prompt and see if it says fl0 was marked for autostart. If it wasn't, something is corrupting your EEPROM on power-down or power-up. Let me know what you find out.
7. js, now that you have it running, you can do anything you want with it!
Actually, I intend to provide support for PWM and servos soon, so robotics is right around the corner.
8. jrseattle, sparrow2, and others,
Yes, it would be fun to have a Puget Sound AVR gathering. If you guys want to hook up for pizza and geek-talk some evening, let me know.
So many features to add, so little time...
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 26, 2008 - 05:33 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
Oops, sorry for the double-post!
sparrow2, I reread your post and realized my reply wasn't actually on-target; sorry about that. I'll try again.
I borrowed Gordon Doughman's original 'hc11 design for the interpreter. His design has some excellent features that I wanted to preserve, one of which is that his interpreter can recover the original source line intact, complete with all embedded spaces. The design you promote is much faster and smaller, but is one-way. I had compelling reasons to go with Gordon's design, but I certainly considered what you were proposing. Thanks for the comment!
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 26, 2008 - 07:15 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
For some reason I had missed that you use a specific BASIC I was thinking that you was making your own.(I will read the manual but I'm at work now).
Now I see why you need more RAM
To speed it up you can do like the spectrum BASIC, store both (1000 is stored as 1000 and $03E8 you just don't see it).
Are there any way I can run this from the simulator?
In the 80's you could buy a 8052 with BASIC in the ROM (I don't think they made it for the CMOS). Maybe some ideas from there could help.
You could make a token for 2 spaces 3...... (or a token that take a byte for the number of spaces.(I asume you want this for indentation).
Jens
Edit: so does this BASIC run a true ASCII text or is it just stored so you can make a 100% correct LIST?
make it a 8052 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 26, 2008 - 11:13 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
|
sparrow2 wrote:
For some reason I had missed that you use a specific BASIC I was thinking that you was making your own.(I will read the manual but I'm at work now).
Now I see why you need more RAM
To speed it up you can do like the spectrum BASIC, store both (1000 is stored as 1000 and $03E8 you just don't see it).
Are there any way I can run this from the simulator?
In the 80's you could buy a 8052 with BASIC in the ROM (I don't think they made it for the CMOS). Maybe some ideas from there could help.
You could make a token for 2 spaces 3...... (or a token that take a byte for the number of spaces.(I asume you want this for indentation).
Jens
Edit: so does this BASIC run a true ASCII text or is it just stored so you can make a 100% correct LIST?
make it a 8052
Actually, this is my own Basic. It is based on Gordon Doughman's BASIC11, but it is all my own code. I translated Gordon's 68hc11 assembly language source into ANSI C, then compiled the C for Windows and also for the AVR.
How the interpreter stores each source line of your program internally was already mapped out by Gordon and I haven't changed it (much). Gordon's internal storage scheme makes it easy to convert back and forth between ASCII text and the corresponding pcodes. Keeping the flexibility, however, means that the pcode scheme isn't quite as efficient as it might be. Still, I want the user (typically a newbie) to be able to see the source faithfully reproduced with a LIST command, so I kept Gordon's scheme.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 27, 2008 - 12:39 AM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
here's some of the keywords/command words from the original. Maybe it would be easy to edit this and delete those that are N/A and those that are omitted, etc.
Code:
"LIST"
"RUN"
"NEW"
"CONT"
"CLEAR"
"ESAVE"
"ELOAD"
"LLIST"
"AUTOST"
"NOAUTO"
"FREE"
"DATA" xdata();
"LET" xlet();
"READ" xread();
"RESTORE" xrestore();
"GOSUB" xgosub();
"GOTO" xgoto();
"ON" xon();
"RETURN" xreturn();
"IF" xif();
"INPUT" xinput();
"PRINT" xprint();
"FOR" xfor();
"NEXT" xnext();
"STOP" xstop();
"ENDWH" xendwh();
"END" xend();
"REM" xrem();
*/*"SWAP" xswap();*/
"TRON" xtron();
"TROFF" xtroff();
"WHILE" xwhile();
TOKTBL: EQU *
FCB LETTOK
FDB LET
FCB READTOK
FDB READ
FCB RESTRTOK
FDB RESTORE
FCB GOSUBTOK
FDB GOSUB
FCB GOTOTOK
FDB GOTO
FCB ONTOK
FDB ON
FCB RETNTOK
FDB RETURN
FCB IFTOK
FDB IIF
FCB THENTOK
FDB THENS
FCB ELSETOK
FDB ELSES
FCB INPUTTOK
FDB INPUT
FCB PRINTTOK
FDB PRINT
FCB FORTOK
FDB FOR
FCB NEXTTOK
FDB NEXT
FCB STOPTOK
FDB STOPSS
FCB ENDTOK
FDB ENDS
FCB TRONTOK
FDB TRON
FCB TROFFTOK
FDB TROFF
FCB WHILETOK
FDB WHILE
FCB ENDWHTOK
FDB ENDWH
FCB STEPTOK
FDB STEP
FCB TOTOK
FDB TO
FCB COMMATOK
FDB COMMAC
FCB SEMITOK
FDB SEMIC
FCB MEOLTOK
FDB COLLINC
FCB IMLETTOK
FDB IMLET
FCB POKETOK
FDB POKE
FCB EQUALTOK
FDB EQ
FCB OPARNTOK
FDB OPARN
FCB CPARNTOK
FDB CPARN
FCB ANDTOK
FDB ANDS
FCB ORTOK
FDB ORS
FCB EORTOK
FDB EORS
FCB LTTOK
FDB LT
FCB GTTOK
FDB GT
FCB LTEQTOK
FDB LTEQ
FCB GTEQTOK
FDB GTEQ
FCB EQTOK
FDB EQ
FCB NOTEQTOK
FDB NOTEQ
FCB PLUSTOK
FDB PLUS
FCB MINUSTOK
FDB MINUS
FCB MULTTOK
FDB MULT
FCB DIVTOK
FDB DIV
FCB MODTOK
FDB MODS
FCB NOTTOK
FDB NOTS
FCB RTIMETOK
FDB RTIMES
FCB NEGTOK
FDB NEGS
FCB SSCNTOK
FDB SPACE
FCB DIMTOK
FDB DIM
FCB EEPTOK
FDB EEP
FCB PORTATOK
FDB PORTA
FCB PORTBTOK
FDB PORTB
FCB PORTCTOK
FDB PORTC
FCB PORTDTOK
FDB PORTD
FCB PNUMTOK
FDB POUNDSGN
FCB INBYTTOK
FDB INBYTES
FCB TIMETOK
FDB TIME
FCB ONTIMTOK
FDB ONTIME
FCB ONIRQTOK
FDB ONIRQ
FCB RETITOK
FDB RETI
FCB PACCTOK
FDB PACC
FCB ONPACTOK
FDB ONPACC
FCB SLEEPTOK
FDB SLEEP
LFUNCTBL: EQU *
FCB FDIVTOK
FDB FDIVS
FCB CHRTOK
FDB CHRS
FCB ADCTOK
FDB ADCS
FCB ABSTOK
FDB ABS
FCB RNDTOK
FDB RND
FCB SGNTOK
FDB SGN
FCB TABTOK
FDB TABS
FCB CALLTOK
FDB CALL
FCB PEEKTOK
FDB PEEK
FCB FEEPTOK
FDB EEP
FCB HEXTOK
FDB HEX
FCB FPRTATOK
FDB PORTA
FCB FPRTBTOK
FDB PORTB
FCB FPRTCTOK
FDB PORTC
FCB FPRTDTOK
FDB PORTD
FCB FPRTETOK
FDB PORTE
FCB FTIMETOK
FDB TIME
FCB HEX2TOK
FDB HEX2
FCB FPACCTOK
FDB PACC
|
|
|
| |
|
|
|
|
|
|
Posted: Sep 27, 2008 - 08:21 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
From the above list, KLBasic supports the following commands:
LIST
RUN
NEW
CONT
AUTOST (but with changes, see below)
FREE
NOAUTO has been replaced; AUTOST now supports the following variations:
AUTOST show state of autostart
AUTOST OFF autostart disabled
AUTOST EE autostart program in EEPROM
AUTOST FL0 autostart program in flash file 0
AUTOST FL1 autostart program in flash file 1
AUTOST FL2 autostart program in flash file 2
ESAVE and ELOAD replaced by LOAD and SAVE.
SAVE has the following arguments:
SAVE EE save program to EEPROM
SAVE FL0 save program to flash file 0
SAVE FL1 save program to flash file 0
SAVE FL2 save program to flash file 2
LOAD has following arguments:
LOAD EE load program from EEPROM
LOAD FL0 load program from flash file 0
LOAD FL1 load program from flash file 1
LOAD FL2 load program from flash file 2
KLBasic supports the following functions and statements:
DATA
LET (implied)
READ
RESTORE
GOSUB
GOTO
RETURN
IF
INPUT (with optional quoted prompt string)
PRINT (also supports ? as quick-print)
FOR
NEXT
STOP
WHILE
ENDWH
REM (also supports ' as a remark char)
TRON
TROFF
END
: (separates to statements in same line)
AND
OR
NOT
XOR
MOD
+, -, *, /
<, >, =, <>, <=, >=
DIM (declares size of arrays)
RND(N) returns random 32-bit int <N
RND(0) reseeds random number generator
This should get you going...
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 29, 2008 - 12:34 AM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
I have version 0.5 of KLBasic ready to go. LFs are now ignored from the console.
Note that there are two files in this release, one for the 'mega128 and one for the 'mega1284p. The 'mega1284p version supports 11K of code space and 1K each of variable and array space. I'm running my '1284p at 20MHz so the benchmark01 program shows 50K empty loops in 0.811 seconds. |
|
|
| |
|
|
|
|
|
|
Posted: Sep 29, 2008 - 07:15 AM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
| the mega128 is still 16MHz? |
|
|
| |
|
|
|
|
|
|
Posted: Sep 29, 2008 - 03:41 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
|
stevech wrote:
the mega128 is still 16MHz?
Sorry for the omission. Yes, the 'mega128 is 16 MHz, the 'mega1284p is 20 MHz.
I'm open to suggestions on how to support a variety of crystal speeds across several MCU types. Besides the problem with console baud rates, changing the crystal impacts the down-counting timers and the uptime variable.
As it is, I'm keeping three different MCU boards laying around; I really don't want to have to switch crystals so I can test-drive versions with different clocks.
I also don't want to just turn loose a .hex file that I haven't tested because follow-on support becomes a nightmare if I don't have a suitable platform.
Any thoughts?
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Sep 29, 2008 - 03:52 PM |
|
Joined: Apr 01, 2004
Posts: 3812
Location: New Mexico
|
|
|
klaxon44 wrote:
Any thoughts?
External jumpers to set the program's knowledge of clock speed at initialization time? Open-source project so that people can compile the source based on their own CPU speed? Defined EEPROM location where end-users can program their own clock speed identifier? [auto baud rate detection has already been discussed] |
_________________
Kevin Rosenberg
http://b9.com
http://kevin.hypershots.com
|
| |
|
|
|
|
|
|
Posted: Sep 29, 2008 - 03:56 PM |
|
Joined: Sep 04, 2002
Posts: 21396
Location: Orlando Florida
|
|
| How about a 'limp home mode' that boots at internal 1mhz and 1200 baud until you pick an xtal speed somehow? What percent is the flash filled up? Need a collaborator to put nokia graphics module or a fat on an sd card module in there? |
_________________
Imagecraft compiler user
|
| |
|
|
|
|
|
|
Posted: Sep 29, 2008 - 06:29 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
Or have the first addr in the eeprom (or something like that) deside the speed, or even put it in a well defined byte in the hex file.
But why not have the normal way where you send a space and the taget make the speed calc?
Jens |
|
|
| |
|
|
|
|
|
|
Posted: Sep 30, 2008 - 01:21 AM |
|
Joined: Sep 04, 2002
Posts: 21396
Location: Orlando Florida
|
|
| Thats what they cal 'autobaud'... if you know the char he's typing, and the xtal speed, you can either measure the bitwidth or look up the char you got in a table of different baudrates. I think the bet's off if the program doesnt know the xtal freq. But I actually thought the idea about having the initial turnon using 1mhz and 1200 baud was pretty clever. That's why I'm mentioning it again. Brazen hubris. |
_________________
Imagecraft compiler user
|
| |
|
|
|
|
|
|
Posted: Sep 30, 2008 - 07:02 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
Can the 1280 switch from intern osc to extern on the fly ? (I don't think so)
Jens |
|
|
| |
|
|
|
|
|
|
Posted: Sep 30, 2008 - 07:29 PM |
|
Joined: Jul 18, 2005
Posts: 62944
Location: (using avr-gcc in) Finchingfield, Essex, England
|
|
|
Quote:
Can the 1280 switch from intern osc to extern on the fly ? (I don't think so)
Nope (so you were right to think not)
I'm intrigued to know why on earth one would want to do that anyway? |
_________________
|
| |
|
|
|
|
|
|
Posted: Sep 30, 2008 - 08:37 PM |
|
Joined: Oct 07, 2002
Posts: 2058
Location: Denmark
|
|
How else would you start at 1MHz so the baud rate is fixed, when you want to run 16 or 20 MHz.
Jens |
|
|
| |
|
|
|
|
|
|
Posted: Sep 30, 2008 - 09:56 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
|
bobgardner wrote:
How about a 'limp home mode' that boots at internal 1mhz and 1200 baud until you pick an xtal speed somehow? What percent is the flash filled up? Need a collaborator to put nokia graphics module or a fat on an sd card module in there?
I am DEFINETLY looking to put SD/FAT on a '1284p! That would let me open up datalogging, text files, and other goodness.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Nov 19, 2008 - 06:54 PM |
|
Joined: Oct 21, 2008
Posts: 4
|
|
I loaded the mega128 .hex file on my home-made "wiring.org" board.This interpreter works great.With 8 LED's connected to port "c" I was able to test the port with a small program.
10 ddrc=$ff
20 for x=0 to 255
30 portc=x
40 for t=1 to 2000
50 next t
60 print x;
70 print " ";
80 next x
RUN
Thanks |
Last edited by posix1965 on Nov 20, 2008 - 05:05 PM; edited 1 time in total
|
| |
|
|
|
|
|
|
Posted: Nov 19, 2008 - 08:32 PM |
|
Joined: Mar 07, 2001
Posts: 2376
Location: Stockholm, Sweden
|
|
|
farang wrote:
While not a serious replacement for C or Basic compilers (remember Forth?), a basic interpreter is an excellent tool for learning programming concepts, and for debugging hardware -the peeks and pokes and input and output commands are excellent ways to tickle port pins or look at them.
Isn't that what the Dragon/JTAGICE is for? Which you just used to downoad the Basic Interpreter with? 
In ye olde days, yes, it was great for debugging and testing, but today?
Especially not a closed source one. There are plenty of open source variants around.
And who hasn't written one themselves? |
_________________
/Jesper
http://www.yampp.com
The quick black AVR jumped over the lazy PIC.
What boots up, must come down.
|
| |
|
|
|
|
|
|
Posted: Nov 19, 2008 - 08:37 PM |
|
Joined: Mar 27, 2002
Posts: 18757
Location: Lund, Sweden
|
|
|
Quote:
And who hasn't written one themselves?
Bill?  |
|
|
| |
|
|
|
|
|
|
Posted: Nov 19, 2008 - 09:53 PM |
|
Joined: Oct 21, 2008
Posts: 4
|
|
All my .hex file are downloaded using Avrdude on a Linux terminal
emulator,and while I still like programming in C it sure beats getting an output without having to program the AVR's flash.
I can't wait until PWM and other IN/OUTS are available using the interpreter. |
|
|
| |
|
|
|
|
|
|
Posted: Nov 19, 2008 - 10:13 PM |
|
Joined: Aug 13, 2006
Posts: 6764
Location: Bellingham, WA - USA
|
|
|
Quote:
Bill?
Since this funny came from you, Johan:
A thought in young Billie's head swirled
And finally it danced and it whirled
"A BASIC I'll claim
Without the least shame
And with it I'll soon rule the world." |
_________________
Chuck Baird
"It's better to catch the trapeze than test the safety net" -- RPi book
http://www.cbaird.org
|
| |
|
|
|
|
|
|
Posted: Nov 19, 2008 - 10:29 PM |
|
Joined: Jun 04, 2007
Posts: 1075
Location: Cincinnati, Ohio - USA
|
|
| I think you've created a monster Johan! |
|
|
| |
|
|
|
|
|
|
Posted: Jan 02, 2009 - 04:54 PM |
|
Joined: Jul 08, 2002
Posts: 4
|
|
I try this nice piece of code.
But the most comun AVR is the MEGA32 I guess.
Can you supply a version for the AT MEGA 32 or the AT MEGA 644 p ?
Thanks to submit your work.
Have an happy new Year
Georges |
|
|
| |
|
|
|
|
|
|
Posted: Jan 02, 2009 - 05:14 PM |
|
Joined: Mar 27, 2002
Posts: 18757
Location: Lund, Sweden
|
|
|
Quote:
But the most comun AVR is the MEGA32 I guess.
Do you have any numbers to put behind that? Or is it really a wild guess / shot in the dark? |
|
|
| |
|
|
|
|
|
|
Posted: Jan 02, 2009 - 07:14 PM |
|
Joined: May 30, 2004
Posts: 8143
Location: Cincinnati, Ohio
|
|
|
JohanEkdahl wrote:
Quote:
But the most comun AVR is the MEGA32 I guess.
Do you have any numbers to put behind that? Or is it really a wild guess / shot in the dark?
If you go by the numbers in my parts bin, I'd have to agree that the Mega32 was the more popular AVR, at one point in time.
But now I seem to have nearly that in each of the Mega168, Mega324, Mega644 & Tiny2313 controllers - all in TQFP packaging, of course. |
_________________
Carl W. Livingston, KC5OTL
microcarl@roadrunner.com
"There are only two ways to sleep well at night... be ignorant or be prepared."
The original Dragon Slayer !
Long live the AVR!!!
|
| |
|
|
|
|
|
|
Posted: Jan 02, 2009 - 10:16 PM |
|
Joined: Apr 25, 2004
Posts: 3817
Location: Denmark
|
|
|
|
|
|
|
|
Posted: Jan 02, 2009 - 10:57 PM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
|
Bingo600 wrote:
If some kind soul would translate the ChipBasic2 PDF user docs to english, we'd be so grateful! |
|
|
| |
|
|
|
|
|
|
Posted: Jan 02, 2009 - 11:59 PM |
|
Joined: Oct 21, 2008
Posts: 4
|
|
[f some kind soul would translate the ChipBasic2 PDF user docs to english, we'd be so grateful!]
And only the chipbasic2 for the m644 keyboard layout is switchable from German to US English. |
|
|
| |
|
|
|
|
|
|
Posted: May 16, 2010 - 12:01 AM |
|
Joined: May 14, 2010
Posts: 1
|
|
Hi,
Is there a manual available for this yet?
Regards,
Gerard |
|
|
| |
|
|
|
|
|
|
Posted: Jun 11, 2010 - 03:12 AM |
|
Joined: Oct 21, 2008
Posts: 4
|
|
|
gerlynch wrote:
Hi,
Is there a manual available for this yet?
Regards,
Gerard
I think this is DEAD. |
|
|
| |
|
|
|
|
|
|
Posted: Apr 15, 2012 - 07:19 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
Not dead; I just forgot to check this thread. :-)
Check my webpage at: www.seanet.com/~karllunt
for the latest release 15 Apr 2012. Sorry, no manual available yet, but I do provide a fairly large test program that shows how to use a lot of the features.
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Apr 17, 2012 - 02:55 AM |
|
Joined: Dec 18, 2001
Posts: 4780
|
|
|
klaxon44 wrote:
Not dead; I just forgot to check this thread.
Check my webpage at: www.seanet.com/~karllunt
for the latest release 15 Apr 2012. Sorry, no manual available yet, but I do provide a fairly large test program that shows how to use a lot of the features.
klaxon44
Just 2 more features and I'd give it a good go...
1. Renumber [by n]
2. labels, so one can gosub alpha
keep an eye out on ZBasic.net too - for new things short term. that's the best structured Basic for 8 bitters. |
|
|
| |
|
|
|
|
|
|
Posted: Apr 17, 2012 - 06:22 PM |
|
Joined: Jul 08, 2002
Posts: 4
|
|
Other ideas
adc()
PWM()
I2C
ON INT gosub...
Nice work
Georges |
|
|
| |
|
|
|
|
|
|
Posted: Apr 18, 2012 - 09:07 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
stevenc, I checked out ZBasic's pages. Looks excellent! If I understand correctly, this is a PC-based compiler that produces either interpreted or native code for an AVR target, right? Very much like my older SBasic for the 68hc11/12/68k. With KLB, I wanted an on-target interpreter so you can develop and debug locally.
I'm open to suggestions on syntax for the labels, as I've given that idea a lot of thought. How about these possibilities:
Code:
100 gosub ReadADC
.
.
3000 label ReadADC
or
3000 ReadADC::
or
3000 :ReadADC
I would like to not use SUBROUTINE as part of this solution, as I may actually get ambitious enough one day to include support for true subroutines and I don't want to create a problem between SUBROUTINE as a target for a goto/gosub and a true SUBROUTINE.
georges, thanks for your suggestions. I've got them in the queue...
klaxon44 |
|
|
| |
|
|
|
|
|
|
Posted: Apr 18, 2012 - 09:17 PM |
|
Joined: Dec 11, 2007
Posts: 6982
Location: Cleveland, OH
|
|
I wondered why the double colon?
Isn't one colon, after the label text, the "standard"?
If one has line numbers then isn't:
1000 goto 5000
.
.
.
5000 Instruction
5010 Instruction
5020 Instruction
.
.
.
also a valid construct?
But, the real question is this:
What do Line Numbers give you that the software can't figure out on its own without them?
JC |
|
|
| |
|
|
|
|
|
|
Posted: Apr 19, 2012 - 03:39 AM |
|
Joined: Mar 12, 2004
Posts: 1207
Location: Portland, OR, USA
|
|
|
klaxon44 wrote:
If I understand correctly, this is a PC-based compiler that produces either interpreted or native code for an AVR target, right?
Partially. The free ZBasic compiler generates either VM code or native object code for the several AVR-based ZX devices of the VM (interpreted) and native types, respectively. With the addition of a ZBasic Generic Target License, the same compiler is able to compile a ZBasic application to native object code that can be downloaded to an ordinary AVR. (Note, however, that not all AVR chips are supported - but over 100 different chips are supported.)
Downloading a compiled ZBasic application to a generic AVR can be achieved via ICSP Flash programming (requires a programmer and downloading application, e.g. avrdude), via a ZBasic-compatible bootloader (source code for which is provided), or via another bootloader of your choice (requires a compatible downloading application, e.g. FLIP). |
_________________
Don Kinzer
ZBasic Microcontrollers
http://www.zbasic.net
|
| |
|
|
|
|
|
|
Posted: Apr 19, 2012 - 11:52 PM |
|
Joined: Sep 22, 2008
Posts: 47
Location: Bothell, WA
|
|
Doc,
One colon following the label text is common, but colon is also the mid-line terminator, as in:
100 a = 4 : b = 3
It would be trivial to have the parser differentiate between the two uses of colon, but I was wondering if anyone had a better idea.
Having a label as a jump target helps minimize the chance of an editing error going undetected. A "gosub 160" when you meant "gosub 150" will still run without complaint (assuming line 160 exists). A "gosub ReadACD" when you meant "gosub ReadADC" will throw a parsing error since the label doesn't exist.
Another benefit comes with program structure. If you have a subroutine in lines 3000-3999 and you have to move it to lines 8000-8999, you won't also have to search and change all instances of 3000 in your mainline code. You can just leave the invocations as ReadADC and the parser will be happy.
The larger issue, the need for line numbers, arose because there simply isn't enough RAM in most of these AVRs to support a RAM editing buffer. For example, if a FOR-NEXT loop spans 20 lines, the editor would have to hold all of those lines so the parser could confirm that the FOR had a matching NEXT.
One workaround would be some kind of standardized, fast-access, external memory, such as an SPI-based FRAM. Way cool, but then anyone wanting to use the Basic program would have to add the device.
klaxon44 |
|
|
| |
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
