Commodore VIC20: my original 1984 machine language game program disassembled from casette tape
QUIKMAN: 1984 revisited by Robert Hurst on October 27, 2008
A couple of years ago, I was rejoined with my old Commodore VIC20 software library my brother had kept. Some of which were programs I wrote and saved using ordinary audio cassette tape. I decided to try and resurrect the data, because I still have my old VIC20 and datasette drive. But if the data successfully loaded into the computer’s memory, how could I then transfer it over to a modern PC? This problem cannot be new to me, so after a few google searches, I learned of a PC serial cable and some DOS software that would allow me to hook up and use a Commodore floppy drive… but not tape.
In steps ebay. I was able to purchase a new-in-the-box VIC 1540 diskette drive for $15 shipped. I then got pillaged another $10 for a box of 5-1/4″ floppy disks. Fortunately back then, I practiced the good discipline of recording on both sides of each tape, as well as to keep a master copy of all things relevant on an additional tape. So with a freshly cleaned datasette head and a LOT of load attempts, I was able to retrieve all of my saved programs from the 22-year aged tapes.
I forgot about the “fun” it was to format a 170kb floppy disk. Compared to cassette tape, floppies were amazingly faster. Today, you can download a copy of it faster than you can type “RUN” on the VIC’s keyboard. Seriously though, using the tapes and floppies was like experiencing it new all over again. And that was kind of fun, though I am glad not to be fussing over such clumsy media with severely limited storage capacity.
I have been able to enjoy the result of this tape librarian nightmare through the use of machine emulation software, specifically from the VICE Team. But a funny thing struck me this past week — one of my programs is a game I wrote in 1984 called QUIKMAN. I named it QUIK instead of you-know-who, because earlier that year I wrote my first fully machine language program dubbed QUIKVIC — quick as in fast, which it really was on a 1mHz 8-bit 6502 CPU. As it went for me back then, I had this one final week off during the last of my college days, and I decided to spend it writing this game. I believed then that there would never be another opportunity for me to create this game, because I was grooming to be a professional data processing programmer using mainframes — my dream of being an arcade game programmer would die, but I needed to try and do this one last time.
I abandoned both my girlfriend and bathing that week, and spent 20-hour days in front of the parlor’s Zenith color TV with my little VIC20 and machine language monitor cartridge. The result of the game came out just fine for something that ultimately loads & runs on a machine with only 3.5kb of memory and 8-colors. As a matter of opinion, my recreation of this arcade megahit is far, FAR superior than what the licensed owners produced for the home computer market of that day. Still, I have always felt I could have done better… If only I had the time, and perhaps even the tools. And now after 24-years, that feeling of an incomplete job has resurfaced.
What triggered that gut reaction was my accidental discovery of a software project on cc65.org. It stirred up fond memories of my first C compiler, as it was also for the 6502 CPU powering the mighty Commodore 128. But this was not that product of that day. However, it sports a nifty 6502 assembler with preset configurations to compile for Commodore 8-bit computers, including the VIC20! I used to own Merlin 128, too, so I had some pretty high expectations from this tool.
After some light reading of its documentation, I became convinced that I could resurrect my QUIKMAN code into an original assembler source format that could be recompiled back into its original binary format. Turning back to VICE, I loaded QUIKMAN, virtually attached the VICMON cartridge, and had it virtually print (to an ASCII text file) a disassembled listing of its machine code and data. Here is a copy of that listing.
Over the past 5 days, I have been massaging that listing into newly-formatted
assembler source, worthy of today’s coding standards. The goal was to produce
an assembler source version that would compile into a binary that was EXACTLY
the same as the originally hand-coded machine language version. After my first
successful pass at compiling, I simply could not wait to look for deltas — I
had to boot it up and see if the program ran. Naturally, I was disappointed
when the screen turned blue and did nothing. I found the first “bug” and fixed
it, and to my surprise and delight, a version of QUIKMAN was up and running.
Way too cool!!
I then had the chore to compare the new binary against the old one. It is
really important to complete the first objective in making an assembler source
that would compile exactly as the original. To accomplish this without too
much effort, I turned to the use of two command-line tools: hexdump and meld.
By issuing:
hexdump -C quikman.p00 > quikman.old hexdump -C quikman.prg > quikman.new
I could then compare the two outputs with this graphical diff view:
meld quikman.old quikman.new
It highlighted just a few differences, which had no real adverse affects on the program functionality, but I wanted it to be precisely the same. After a few more edits, validated by a clean meld view, the assembler source is now complete.
Now I wonder how many more days I’ll go without bathing until I figure I am done with its next revision … ?