don't we still need to pass --enable-impeller sometimes? I'm having a bit of a hard time wrapping my head around two independent flags 🙃

This function creates the flags for iOS and iOS simulator. The --enable-impeller(=true) flag is always a no-op there after the engine flag flip. It can't override the Info.plist setting. When there is no Info.plist setting either way, the engine flag --enable-impeller=false will disable Impeller on iOS and iOS sim.

After this change, when --enable-impeller is passed on Android, --enable-impeller will still be passed to the engine.

But we're not changing the info.plist value, just the default if not specified right? Can we update --no-enable-impeller to pass --enable-impeller=false?

What I was having trouble figuring out was how to handle the case where the --enable-impeller flag is absent on the command line. The default value that we'd like the flag to have differs depending on the target device, so DebuggingOptions would need to know both the value of the flag, and whether the flag was present or absent on the command line. Do you have thoughts on whether that's preferable to the --disable-impeller flag?

I don't think debugging options actually needs to know, the only difference for the tooling is what shaders are generated and we've so far worked around that by just generating both variants

We could instead update this logic to always generate both variants for iOS and call it good?

we already do this: https://github.com/flutter/flutter/blob/master/packages/flutter_tools/lib/src/build_system/targets/shader_compiler.dart#L57

Switched to an enum as discussed offline. PTAL.