4 ms·
If you can get a simulator up and running, Macintosh Garden has the originals available (feeling this is ethically OK to link now that this is open): http://ma
by matheweis 11y ago
If you can get a simulator up and running, Macintosh Garden has the originals available (feeling this is ethically OK to link now that this is open):
http://macintoshgarden.org/games/glider-312 http://macintoshgarden.org/games/glider-312 (very early black and white version)
http://macintoshgarden.org/games/glider-40 http://macintoshgarden.org/games/glider-40
http://macintoshgarden.org/games/glider-pro http://macintoshgarden.org/games/glider-pro
- tbrock 11y ago"You can’t open the application “Glider Pro.app” because PowerPC applications are no longer supported." :(
- pmarreck 11y agoYeah, gotta use an emulator of some sort I guess. I really wish Rosetta was kept around by Apple.
- duskwuff 11y agoRosetta depended on Apple maintaining PowerPC builds of all of their frameworks. I have to imagine that was a huge drag for their developers; there's no way they would have been happy about maintaining that.
- wtallis 11y agoJust making sure that most of userspace still compiles for PPC is probably worth the effort for Apple, to ensure the code remains portable. Maintaining the platform-specific intricacies of the language runtime and kernel-space stuff would be a drag, and it's probably difficult to maintain the hardware necessary to test it all on, but that may not be required to keep Rosetta working.
- mrpippy 11y agoI would bet $$$ that they compile and run the whole OS on ARM, much easier than PPC since so much is shared with iOS
- duskwuff 11y agoI would be shocked if they didn't! They've already got an ARM Darwin kernel (for iOS), and they already cross-compile a bunch of the iOS frameworks to x86 for the simulator. Cross-compiling OS X to ARM is a natural next step.
- pmarreck 11y agoThe game company Blizzard has (for as long as they have existed) maintained multiple build paths for their code (Mac and PC). I can't find the argument now (of course) but their lead dev argued that forcing themselves to do that meant they ended up making MUCH better code as a side effect, because it HAD to be modular/agnostic, which also made it more resilient, less buggy, more easily testable, more modular, etc. Sure, you'd have to deal with endian-ness differences everywhere... or used to, when PPC was still a thing