AFAIK this is incorrect the simulator runs x86/x86-64 binaries and links against x86/x86-64 versions of Apple's frameworks (including UIKit). There is no translation of syscalls happening, the instruction for a syscall, e.g SVC 0x80 on ARM wouldn't be happening in your binary but in a framework your binary (indirectly) links to.
Also the simulator is basically an x86 version of iOS, look inside for yourself and see. Someone correct me here but I don't see why it would be impossible to for apple to ship a simulator for Windows, n.b. how MS has shipped the Linux ABI translation.
Re:Apple Starts to Woo Its App Developers
by kitsunesoba on Saturday June 11, 2016 @03:49AM
(#11882555)
The simulator doesn't run a full x86 version of iOS; it runs an x86 version of the iOS userland on top of the host kernel (Darwin) and components shared between OS X and iOS.
A port of the simulator would be possible but it'd be a good deal heavier than its OS X counterpart since it'd be running the entire OS instead of just the topmost layer.
Re:Apple Starts to Woo Its App Developers
by sjtgraham on Saturday June 11, 2016 @06:15AM
(#11882894)
Both platforms share the same core OS (Darwin) and Kernel (XNU), so when I say the simulator is basically an x86 iOS I think that's fair.
You're again missing the point that that would be heavier and so perhaps not as good for developers that choose to use macOS.
The question is not whether "the simulator is basically an x86 iOS". The point is that requiring macOS to build iOS apps is not an arbitrary restriction — it's a tradeoff. You can argue for or against Apple's decision, but you can't argue that it's an arbitrary restriction on Apple's part.
Why is the simulator not simply a VM loaded with iOS, like with Android, or Windows Phone, or most of the other targets that you can cross-compile and test for from Windows or Linux?
There are significant downsides. You can argue that the upside is still worth it, but you can't dismiss the downsides as "completely arbitrary".