I've decided to take on the task of porting the leaked NESticle source to modern platforms. The end product isn't intended to be practical. If one's looking for an NES emulator, there are plenty of modern options available. This is purely for fun (and maybe profit); it is indeed, a labor of love.

At the moment, the plan is to post an article for each significant point of the project's development. I am also live streaming the development on Twitch. However, I don't actively engage with the audience much on Twitch and unlike these posts, the flow isn't as digestible for viewers. You are however, welcome to watch me struggle in real-time.

Also, do not expect the posts to remain in sync with the streams. The articles will lag far behind.

With respect to end goals, I am at the moment bias to preserving as much of the original NESticle content and methodology as possible; at least until the first working build/release. I imagine there are many improvements to be made. The code-base is a mixture of C++ and 32 bit x86 assembly. One obvious point of improvement would be removing the low level assembly code and make the code-base more portable. I'd love to see NESticle running on my Raspberry Pi. However, for reasons of strategy and historical preservation, I'm keeping the lack of portability and other inherit faults by avoiding implementing initial improvements.

I have already ported a chunk of the assembly code. For now, you should expect the initial releases to be 32bit x86 Linux executables.

Tackling this project means moving through many unknowns. It's been a long time since NESticle's source code was updated. The C++ standards have changed; the compilers have changed; the operating systems have changed. We need to methodologically break down the problem and keep the unknown variables down to a minimum at each step. Therefore, the first thing to do is simply get each source file to compile. After we accomplished that, we can begin building off of the potentially working code.

But, before we can start compiling, we need the source code which can be downloaded at The Archive. After extracting the zip file, we see this mess of files and directories:

BACK/ COMMON.H CUNT.ZIP DSOUND.LIB FUCK.ZIP GUIROOT.H INPUTW.CPP LOCAL.H MOUSE.CPP NES.CSM NES.EXE NES.MBT NESSOUND.H NESVIDEO.LIB PTRN.ZIP README SCROLLFK.ZIP SOUND.H STDDLG.CPP TYPES.H VOL.H BUILD.CPP CONFIG.CPP CYCLE.ZIP EXAM GUICOLOR.H GUIVOL.H INPUTW.H MAIN.CPP MOUSE.H NES.~DE NES.H NES.MRT NESSRC.ZIP NESVIDEO.OLD R2IMGASM.ASM README~ SFUCK.ZIP SOUND.ZIP SYNC.ZIP USER.LIB VULNS.H BYTEOR~1.H CONFIG.H DD.H FILE.CPP GUI.H INCLUDES.H INSTALL MESSAGE.CPP NAMESERV.H NESDLG.CPP NESH~1.OLD NES.OBR NESTICLE.ICO PATTERN.ZIP R2IMGASM.OBJ RESOURCE.H SHITFUCK.ZIP SPRITE.ASM TILE.ASM UUTIMER.H WIN95.LIB CLIP.H CPU.CPP DDRAW.LIB FILE.H GUI.LIB INPUT.CPP KANJI.H MESSAGE.H NES95.MK NESDOS.MK NES.ICO NES.OLD NESVIDEO.CPP PPU.H R2IMG.CPP ROM.CPP SLIST.CPP SPRITEBG.ASM TILE.OBJ VERSION.H COMMAND.CPP CPU.OLD DISASM.CPP FONT.CPP GUIMENU.H INPUT.H KEYB.H MMC.CPP NES95.WPJ NESDOS.WPJ NES.IDE NESPAL.H NESVIDEO.FUK PROF.CPP R2IMG.H ROM.H SLIST.H SPRITEBG.OBJ TILE.OLD VOLBATCH.H COMMAND.H CRASH.ZIP DLGPOS.H FONT.H GUIRECT.H INPUT.LIB LOADPARM.H MMC.H NES.CPP NES.DSW NES.LIB NESSOUND.CPP NESVIDEO.H PROF.H R2.LIB SCRIPT1.RC SNAPSHOT.CPP SPRITE.OBJ TIMING.H VOL.CPP

Unfortunately, it's not clear what's going to be important. The README is not going to be helpful, because it's a completely unrelated document titled "NetBIOS Security Kit".

To make matters worse, all those zip files are different revisions of the code-base. Thus, we're going to have to decide which revision to port. At the moment, I suspect the newest revision is the best place to start.

The files in the top directory are from April 4^th, 1997. Running the following in my fish shell,

for z in *.ZIP
    unzip -l $z | awk -v z=$z '$2 ~ "199[0-9].*" {print $2, z}'
end | sort -r | uniq

we get the following (truncated) output:

1997-03-31 SOUND.ZIP
1997-03-30 SOUND.ZIP
1997-03-29 SOUND.ZIP
1997-03-28 SOUND.ZIP
1997-03-27 SOUND.ZIP
1997-03-27 CYCLE.ZIP
1997-03-27 CRASH.ZIP
1997-03-26 SOUND.ZIP
1997-03-26 CYCLE.ZIP

# Omitted output

1995-02-06 SOUND.ZIP
1995-02-06 NESSRC.ZIP
1995-01-25 SOUND.ZIP
1995-01-25 NESSRC.ZIP
1994-01-16 SOUND.ZIP
1994-01-16 NESSRC.ZIP

Each line corresponds to a file within one of the zip archives. The left column is the modification or creation date and the right column is the zip archive the file belongs to. The lines are sorted by the date in descending order.

From this output, we can see that SOUND.ZIP is the latest revision from among the zip files. The files in the top directory are newer but after poking around, it looks like the top directory is incomplete. There's missing files such as 6502/6502.CPP which exists in SOUND.ZIP. I'm betting starting with SOUND.ZIP is the safer bet. As we progress, we can always cross compare the different revisions with our current working directory.

Now that we picked which revision of the code-base we're porting, we need to prep the source files for working with on Linux. This code was originally written and built on DOS and Windows 95 with the Watcom C/C++ tool chain; all of which are ancient.

For starters, filenames on the the FAT file system (which is what DOS uses) convert all filenames to uppercase. We're going to need our filenames lowercase. If we don't convert them, the C++ preprocessor won't be able to include any of the header files. Our modern file systems are (usually) case-sensitive. If one of our source files is trying to include NES.H by doing #include "nes.h", the compiler will throw an error because it can't find the file. Therefore, let's wack down our majuscules to minuscules.

# First change directory names to lowercase. 
for f in (find . -t d)
    mv $f (echo $f | tr [:upper:] [:lower:])
end

# Then change filenames to lowercase.
for f in (find . -t f)
    mv $f (echo $f | tr [:upper:] [:lower:])
end

Next, we need to change how newlines are defined. Anyone who has copied a file from Windows over to a Unix-like operating system has encountered this problem. The standard way of putting a newline in a Windows text file is with "\r\n" while on Linux (and Unix) it's "\n". If we take a peek at the first line of main.cpp, we can see the extra "\r" byte for the newline.

$ head -n1 main.cpp | xxd -g 1
00000000: 2f 2f 36 35 30 32 20 6e 65 73 20 65 6d 75 6c 61  //6502 nes emula
00000010: 74 6f 72 20 66 72 6f 6e 74 20 65 6e 64 0d 0a     tor front end..

The two bytes 0d and 0a correspond to "\r" and "\n" respectively. To fix this, we just need to use a tool called dos2unix.

for f in (find . -type f)
    dos2unix $f
end

If we double check the first line of main.cpp, we'll confirm the "\r" is gone. I'll let you do it yourself.

Now that we have the code-base prepped, we can start digging around trying to find a direction of attack. But first, let's commit this into a git repository as our initial starting point.

git init
git add -A
git commit -m init

From this point, we'll begin the actual porting of the code.

1. The leaked NESticle source code.
2. The Story of NESticle, the Ambitious Emulator That Redefined Retro Gaming
3. Complete archive of all NESticle versions.
4. Fish Shell: A smart and user-friendly command line shell.