If you start http://0x10co.de/htg6u it will echo every (ascii) key on the screen, and return does work, too, but if you hit tab, the program will not see it (and switch to the next screen) and instead further keyboard input will be bunched and out of order. All dead keys, arrow keys, PgUp/PgDown and Fn keys have the same effect.
I'd like to tell you which version of firefox I'm running in case the behaviour is browser dependent, but all I can say is that it is one affected by this: https://bugs.launchpad.net/ubuntu/+source/firefox/+bug/920606

does not work: http://0x10co.de/47x8
works: http://0x10co.de/jdyjm

identical code, the only difference is one is using 0x00 for end of string, working copy is using 0x0A

im not entierly sure what is happening but changing the check on line 47 made it all work

Could you add area under code for comments, so we can discuss with author of code, propose alternative solutions, etc.?

For start Intense debate/Disqus/Facebook comments would be enough, but in long run own solution should be baked.

It would be amazing if this worked on the iPad, then I could sit in bed and hack! Currently hitting 'Run' does nothing at all. If it helps, I'll see what I can do to debug it on mine..

The following program fails with the following message: "Assembler error: Label * was not defined (address 5)"

set [0x8000], 0xf041
brk
dat ": X Y * ;"

Did you think about incorporating Notch's emulator here (possibly with some changes, i.e. code loader) instead of making custom one? This would make sense as Notch permits use of his code with or without modifications, unless you take money for it, so...

Of course he should still be asked if he wants it or not.

No idea what's going on here. I'm simply trying to set a font word. It fails when setting to 0xe010. If I set it with a literal value, it works. If I change 0xe010 to anything else, it works.

http://0x10co.de/5j03d

When you run any program that uses keyboard input, and then click somewhere, for example on the screen, input stops functioning.

http://0x10co.de/bootstrap/js/bootstrap-transition.js throws a 404
and so far I can't find any page which doesn't use this file :-)

The virtual LEM1802 does not make characters blink when their bit 7 is set.

Interruptions made by hardware are not working for me, for example keyboard.

Not an issue, But first way that came to mind to contact you. I have really been bothered by accidentally leaving the page and my code being erased. I save eventually, but I really hate when I accidentally loose code. Could you do something about it? Temporarily cache the code or something?

The branching opcodes take one cycle longer to perform if the test fails. When they skip an if instruction, they will skip an additional instruction at the cost of one extra cycle. This lets you easily chain conditionals.

The skipping of an additional if instruction seems to be not implemented yet.

Example http://0x10co.de/b2nh3

if the first parameter is between -1..30 then it breaks "ifn" operation.

Most likely the reason is that b value in the opcode does not have literal range.
In this case assembler should switch a and b values or use next word for b.

Probably other if statements have the same issue

Console window in IE is broken

Could you add "All programs" list with all programs uploaded to 0x10co.de which we can sort by title, author or date of post? Also way to search in code or in name/description/author would be cool.

Of course "private" flag should be added to posting options that if checked make your program not visible in all listings (all/top/recent programs).

Tested in goforth: http://0x10co.de/hiwhx

Goforth expects enter to send the keycode spec'ed in the keyboard spec. 0x10co.de does generate that keycode, but then it also generates an extra 0x13. It looks like the key is being triggered through both the keydown event and the keypress event. The keypress event doesn't check keyMap, which explains why it's not being converted. But either way, it's extra.

Changing line 413 of ui.js to:
if(e.which >= 37 && e.which <= 40 || e.which === 8 || e.which === 13) e.preventDefault();
fixes it.

In Chrome on Windows 7 (64bit) loading the page with the Chrome inspector open shows that https://raw.github.com/mappum/DCPU-16/master/dcpu16.js is throwing a 403.

Possibly github blocking so many requests to a raw file from an external site (I don't know their policy on this)?

Can you make list of examples such as http://0x10co.de/d6ssm that display specific functionality (in this case getting display to work)? This surely would help newbies.

the "DCPU16 Quick Reference" has out-of-date information. O register should be EX, etc. per https://gist.github.com/2341455

not working on chrome build 18.0.1025.168
works on Internet explorer on the same computer

Example: http://0x10co.de/askjz

IFE A, [SP + 1] assembles to
6012 0001
but is should be
6812 0001

Console output works fine in edit mode, but after saving, sometimes nothing will be output to the console (maybe b/c of breaking only after a few thousand cycles?).

examples:
http://0x10co.de/83379 - same as http://dcpubin.com/puTIqosed
http://0x10co.de/vtas - same as http://dcpubin.com/ee2kkbeuJ

I.e. : 0xFFC1 isn't blinking

Hello!

There seems to be a problem with [register + literal].
if the value of a register + the value of literal is below 0xFFFF then it is OK. But in case of overflow it doesn't work

Example

SET A, 0x8010
SET [A + 0xFFFF], 0xF030
SET [A + 0x0000], 0xF030
SET [A + 0x0001], 0xF030

It should display three zeros but it displays only two

I am using Mac OSX Lion. Tried it with Safari, Firefox and Chrome

I can't see anywhere support for palette RAM for LEM1802 display. Can you add it?

I am posting anything that is secured by a password. The address "0x10co.de /?" asks for the password but the address "0x10co.de/fork /?" does not.

the following code:

:q DAT "Hello world!", 0

:start
IFE [q+I], 0
        SET PC, end

SET A, [q+I]
BOR A, 0xC000
SET [0x8000+I], A
ADD I, 1
SET PC, start

:end

works but this:

:q DAT "Enter a number", 0

:start
IFE [q+I], 0
        SET PC, end

SET A, [q+I]
BOR A, 0xC000
SET [0x8000+I], A
ADD I, 1
SET PC, start

:end

only prints "Enter"

Whenever I call HWI repeatedly (in a loop), no matter to which virtual device, the browser window starts lagging to the point of being unusable.

Happens on Chrome, Firefox and Safari in Mac OS X 10.6.8. Also on Windows 7 64bit.

dEgle95 provided an example program:
http://0x10co.de/ymlgk

The I/O specification on 0x10co.de/io states the following:

"The color of the border of the display may also be set by calling a hardware interrupt. The interrupt key is SET_BORDER_COLOR, which is the number 3 on the LEM1802. The interrupt value in the same HRGB color format as cells."

But according to the NE_LEM1802 specifications, the border color works as follows:

"Each color entry has the following bit format (in LSB-0):
0000rrrrggggbbbb
Where r, g, b are the red, green and blue channels. A higher value means a
lighter color."

Which is definitely not the same as with the cells.

The 1.7 spec says that RFI takes one value (though it doesn't say what it does with it). As such, some assemblers require that a value is specified, however yours requires that none is specified. It would be nice if yours allowed an argument to be portable with other assemblers and agree with the spec.

I realize that the argument is not particularly useful, but it should be accepted nonetheless.

• Computerizer

Currently we can only input DEC and HEX numbers into our programs. However sometimes binary is more convenient (like when designing our own font). What I propose is to add "base sufixes" known from most x86 assemblers.

So
XXXXXXXXXXXXXXXXb - binary number
XXXXXd - decimal number
XXXXh - hex number

instead of 0xXXXX for HEX and no suffix for DEC.

Though 0xXXXX/no suffix syntax should remain for compatibility reasons, but should be marked as depracated.

Can we have the console + debugger scroll down to always remain in view when we are typing somewhere
deep inside a code file?

When tweaking code in a large source file, you keep having to scroll up to reach the console for testing. Then having to scroll back down and find the line you were working on.

An extension to this applies to the code running in step mode.
The code editor can jump to the current line, while always having the console/debugger in view.

Chrome, Snow leopard:
Register values in "debug" section don't update while running code in the editor. They do update after code has been submitted.

Reset failed to load new mines.

How about .include instruction? It would automagically, before execution include content of program located at 0x10co.de/, like #include in c.

For example

set a, 5
.include abcde ;inserts program abcde which modifies a register
add 5, 10

will at second line include code of 0x10co.de/abcde.

It would be useful when developing large projects such as operating systems or games.

Not completely sure, but i think that if users hit the keys to hard too often, 0x9010 will show the most recent key. However, if they hit 2 keys before the cycle gets back to looking for keys, it will start to snowball and eventually stop working. I think the easiest fix would be to set it to the next key pressed, not the latest, and then start adding onto the buffer and maxing it out at 16 if the user keeps spamming buttons. If im wrong, feel free to correct me as ill need to fix some of my code ^^.

I am working on creating a compiler which will produce bytecode. 0x10co.de is perfect for testing the assembly, but I have a feeling that compilers will become more prevalent the closer the game comes to completion.

Especially with the server being pounded due to Notch's tweet, it has become clear that private servers are going to be useful.

Obviously this is a node.js server, but with the javascript files named program.js or javascript.js, I honestly don't know where to start.

After following the install procedures, on node app.js execution I get this error:

node.js:201
        throw e; // process.nextTick error, or 'error' event on first tick
              ^
TypeError: Object function createApplication() {
  var app = connect();
  utils.merge(app, proto);
  app.request = { __proto__: req };
  app.response = { __proto__: res };
  app.init();
  return app;
} has no method 'createServer'
    at Object.<anonymous> (/root/0x10code/app.js:19:19)
    at Module._compile (module.js:441:26)
    at Object..js (module.js:459:10)
    at Module.load (module.js:348:31)
    at Function._load (module.js:308:12)
    at Array.0 (module.js:479:10)
    at EventEmitter._tickCallback (node.js:192:40)

I'm running on Ubuntu 11.04 Natty. Any help is appreciated, thanks! I'm just testing some performance optimisations.

http://0x10co.de/ga99b

You have to scroll a long way down to see the actual emulator.

the following code:

SET B, 0xC000
SET A, 0x30
JSR print
BRK

:print
SET PUSH, A
BOR A, B
SET [0x8000+J], A
ADD J, 1
SET A, POP
SET PC, POP

prints the character twice.

tested in firefox and chromium.

Hello, great website.

I was working with a hello world.
When I step through the code, it doesn't seem to refresh the display.
When I click "run", it displays just fine.

If you would like more information, please let me know.
Thanks
Keenan

--- program I was running ---
set i, 0
:loop
set a, [hello_world + i]
ife a, c
set pc, end
bor a, 0x2100
set [0x8000 + i], a
add i, 1
set pc, loop

:end
set pc, end

:hello_world dat "Hello World!", 0

If one has a loop like:

HWN Z

:hw
  HWQ Z
  SUB Z, 1
  SET PC, hw

Not all the hardware devices will be detected by HWQ. For the first iteration in the loop, none of the registers that HWQ sets will actually be set. And only on the second and subsequent iterations will these registers be set.

Right now it's pinned to 100kHz, why not have a little menu where you can set the frequency to your own value or select from a few, like 1kHz, 10, 50, 100, 200 and so on.
Make it easy to test how programs run under-clocked or over-clocked.

The "light black" and "dark white" colors are a bit off. Dark white is greenish, and light black is still black. Demo: http://0x10co.de/b49aw

Can they be added?

HTML-wise, every line, beside line number would have checkbox that indicates BP and when emulator in debug mode run into line with checkbox checked, it'll stop emulation.

When I press apostrophe while I'm running, e.g. DCPU text editor, instead of typing it into editor, in my Firefox 9.0 "Quick search (links only)" bar appears and in result it makes it a bit useless (it's just one key, but still...)

Chrome, Snow Leopard: The space and shift keys fail to work while code is running

Example: http://0x10co.de/xbdgi

set PC, main

:used_memory 
    dat "Used memory: ", 0 
:current_stack 
    dat "Satack: ", 0

:main

    :end
    brk
        set pc, end

This code gives memory dump like that:
0000:_7f81_0012 0055 0073 0065 0064 0020 006d
0008: 0065 006d 006f 0072 0079 003a 0020 0000
0010: 0000 0000[0040]7f81 0012 0000 0000 0000

Why are first two words of the third row zeros?
There should be data for the string "Satack: "