r? @eddyb

(rust_highfive has picked a reviewer for you, use r? to override)

So the libsyntax/entry.rs need a little bit clean up to make those comments to be correct.
But tbh that entry_point_type function serve a different purpose, that is, to mark main-like function #[allow(dead_code)] and to remove start attribute.
But it's not really about entry point, rather, it can just remove all start attribute and make the dead_code lint ignore start and main items, as well as function named "main" in root namespace.
Therefore I suggest it's just improper at first place, and I'll improve it at next PR.

cc #28937

Can this RFC be reconsidered by the lang team?
This RFC adds a strange thing to the language and a special case to the compiler, and the motivation is very weak.

Also, @bombless, could you add a feature gate for this, so even if it's merged now it could be purged later?

Added the feature-gate and the bootstrap hit a problem.. need to fix it first.

Why did this get closed? I'd love to see this implemented; I have a library that would benefit from it.