> I do like the approach, as it makes the (better) fallback-less function
> usage opt-in, which is currently only fixed by coding-standards tooling.
> To be clear, would this lead to a "Fatal error: Uncaught Error: Call
> to undefined function" in case of autoloader failure?
Yes, that would be my expectation. Though I didn't mention it in my first
email, the same approach should work for autoloading constants as well as
If others are in favor, I'd be happy to collaborate on writing an RFC.
I'd need someone more experienced with PHP's internals to help with the