Krzysztof Kluczek, either your source code is buggy and cannot be compiled or i am a donkey using gcc, but i can't be able to compile it, even when i turn off all access checking and warning.

After testing all source code except allegro gl and Krzysztof one, i give my voice to Ryan game.

yeah, kaboom. My bad

Krzysztof: the X11 driver is default on my system, modes that the XWFS driver (does full screen X11 modes) can't handle get done by the XWIN driver (does windowed X11 modes).
and heres how I compiled: g++ file.cpp -o file `allegro-config --libs`. That resulted in an instant crash when run. (SIGSEGV)

edit: ouch, yeah, not checking set_gfx_mode for failure will result in a nice crash... maybe set to GFX_AUTODETECT...

A linux console program "can" be launched from inside X... AFAIK. I just don't have any of the console drivers setup

Ok, I've squeezed my code even more and I was able to make some major improvements! I took some tips from marcello's code(replacing decimal numbers with variables). I replaced a non-portable number with it's proper GFX_OPENGL_WINDOWED. That means this code can now probably run under Linux, although I haven't tested. I'll test as soon as I can for all you Linux people, so expect more news. I also added END_OF_MAIN() for more comptablitiy and the types to a variable and main. The code now compiles under MinGW and MSVC. Attached is the entire package. The source for the 80x25 version, the source for the original source code(before squeezing), the batch file for building under MinGW(probably DJGPP as well), and an already built exe.

I hope this update will allow more people to play my game, especially Linux users, since Linux rules.

Krzysztof Kluczek: Awesome game. I was able to compile it just fine(MinGW32). Very interesting how you used your own sourcecode as the map, heh. I wouldn't have imagined a raycaster would fit in 80x25, but I guess it does!

Well, my program is a bit hacky raycaster since it only draws floors using base/y*cos() and base/y*sin() to determine tile position. If a tile contains a wall the rendering is ended and if it contains a pit rendering floor is delayed. As a result, rendering walls isn`t very much accurate especially at edges (it is possible to look through walls that have one common point, which true raycaster won`t allow).

Pit-falling routine is even more hacky since I`m just copying bottom screen line up the screen using such a speed, that it`s possible to see that something is moving but impossible to see how it is done.

And I used source as a map since it`s the easiest way to be sure that all coins can be reached.

I think I`ll start working over my next 80x25 game. Multiple entries are allowed AFAIK.

btw. Goodbytes, have you posted a news about this contest on Allegro.cc?

Probably MSVC doesn`t understand escapes at the end of some lines (the only explain for unterminated string). Missing semicolons etc. could be results of error recovery after hitting that backslashes. I`ll try to fix that. This program compiles under gcc with only double->int conversion warnings.

(still you could change the file manually and join these escaped lines, just let it read from the old source)

edit: here`s the code without those backslashes:

Nothing special.

you only need to hold the mouse button, it's better that way.

Especially if your mouse pointer is in the middle of the screen. I`ve got 5107 that way.

I compiled it under DJGPP and it complained about return without a value. I fixed it and everything else is ok.

DJGPP libc help says, mem* are in <string.h>
(mem.h still works fine on my compiler)

The only thing I can say, is if you are still running into size problems, couldn't you make your code smaller (on the screen, anyway) by preresolving all of the constants? Also, why not add the comma in your circlefill, rectfill, and text printf functions up above? That way you are only using one comma, instead of one occurance of a comma every function call... (besides, you have tons of room up above by your compiler command things (sorry, can't remember exact term right now, oy...))... heck, you could even through in a semicolon to the beginning of you compiler define, so it looks like...

#define C ;circlefill(screen,

Think of all that wonderful space you are saving! (True, this wouldn't necessaryily work, but it should in theory.)

EDIT--- heck, if compiler defines worked that way, you could do something like

#define db double
#define el else
#define f if (
#define ei else if

but then again, you would be lengthening the program by doing that... too bad you can't stack preprocessor commands...

I'm dumb!... this could get real interesting

You don't have to waste a full line with #defines. Just see how I used them, both #defines are about 80 characters.

It worked like this:

#define A(x,y,z) z?long_func_1(x,y):long_func_2(x,y)

Of course, the extra z param wastes some space, but for me I saved quite some lines with it