I’ve built a game on OpenFL, targeting Windows and OSX. It does some very simple file read/writes to save game data, specifically writing a small json file.
Some of my Mac testers have reported that the game crashes during sections where it’s doing an external load, and sometimes at startup. However, curiously, the issue doesn’t persist when the app binary is run from the desktop of the machine (if it’s placed elsewhere, the crash happens again). This isn’t all users, just a few, I have two test machines here that I’ve never seen crash.
The game contains no fixed/absolute paths whatsoever, and nothing relating to the desktop exists anywhere in the code.
Does anyone have any idea of what would be causing this? There appears to be no pattern in Mac versions, some macs seem to be happy enough and there’s no crashing, and there’s no issues anywhere with the same code running on Windows.
We could, but having a readable save file was easy to work with (it’s a huge game). Either way, the crash also happens at startup when external assets are being loaded, so there’s something going on either way!
EDIT - I had a look at the feedback and we get consistent crashes during the game too when loading external resources. Very strange…
so currently we’re just sending it out as a zipped binary, and eventually it’ll be distributed in probably the same way on itch.io - I could understand if some macs didn’t allow read/write from unsigned random apps, but I’m confused about the desktop part - this definitely fixes it for people…
We ran into similar issues on Windows where our game would work if player’s double clicked the exe directly in the folder, but not when they used a shortcut icon on the desktop. It would just immediately crash.
Turns out it was a very specific issue with loading lime.ndll–if they opened it from the folder, it found the library in the same folder, but apparently looked for it elsewhere if they opened it from the desktop shortcut. We fixed it in our version of OpenFL by modifying the Lime source code. From what I can tell, it has since been fixed on Github.
Maybe something similar is happening with your mac build?
To anyone who come across this issue, it ended up being related to where we were writing data - we switched to embedded external files, and now save our data to the usual applicationsupport location and it went away. Much rejoicing. Thanks, all!
