Like a lot of people in the world, I have not had an outstanding week. I had been dealing with a bout of insomnia that persisted through time changes. That, thankfully, has finally fixed itself this past week but my body is still catching up. Combine that with the general stress of the Corona virus plus some other factors and I have not been in the best of moods. In the interest of trying to improve things, I decided to return to some of my favorite games: LucasArts adventure games.
LucasArts was the game division of LucasFilm. While most people will recognize LucasFilm for Star Wars, LucasArts was responsible for a number of beloved video game series in the adventure game genre. Many of these video games shared a common engine called SCUMM. Fans of these games reverse engineered the engine and started a project called SCUMMVM to play the games on newer platforms using the original data files. I, fortunately, had the original data files that I used on my family computer growing up.
I played through Indiana Jones and the Fate of Atlantis last night. It was both as entertaining and as frustrating as I remember (I remembered most of the puzzles but I hate flying the balloon and driving the submarine. Don’t get me started on the fighting mechanics.). This afternoon, I decided to pull out the first Monkey Island game. Much to my disappointment, the game crashed almost as soon as soon as it started.
A smarter person might have just given up and done something else but I wanted to be a good open source participant and see if I could file a bug. I verified that, yes, it crashed on the latest nightly version as well. I was running this on Mac OSX so I figured I would test it on Linux as well for good measure. It crashed there too. At this point I was already in too deep so I figured I might as well clone the git tree and see if I could get some more information on Linux.
make later, I had the binary running (and crashing!)
in gdb. It appeared to be crashing on a null pointer in sound
initialization. This was in code specific to the mac version so my first
thought was something else changed and missed updating the code. From tracing,
it looked like there was a missing call to
loadInstrument. More tracing
showed that it should have been called in loadMusic
and that it was never entering the loop at all.
idArray.size() was 0. Weird.
I stared trying to look at the resource parsing code when I decided to look
at the binary itself. IT WAS 0 BYTES! The code was checking if the binary
existed (which it did) but it wasn’t pulling anything out of it. The code was
missing an error check somewhere. If I deleted the empty file I got an error
that there would be no music but it didn’t crash.
The mystery of the crash was solved but why did I have a zero byte binary? The other data files seemed to be just fine? I don’t have a 100% answer for sure but I suspect it boils down to how the original Mac OS stored files and my haphazard archive strategy. Files on Mac OS before X had the notion of a Resource Fork and the Data fork. The resource fork was designed to store parts of an application like menu information in a structured format that could be easily changed and edited. I spent a lot of my childhood using ResEdit to edit the resource fork in various games to tweak text strings and graphics. The concepts of ‘forks’ was also very tied to the HFS file system. In order to properly share files across a non HFS file system, you needed to make sure the resource fork was preserved using a file format such as MacBinary, BinHex or Stuffit. I suspect somewhere along the way I failed to transfer the file with the resource fork intact and the data got lost (the move to Dropbox is a likely candidate).
My original plan was to specifically avoid programming for relaxation but this ended up being a nice short distraction which was the important part. Now on to that rubber chicken with a pulley in the middle.