Lint wrong usage of known attributes - spotlight: no_std

[Robbepop]

When compiling a no_std crate it already happened multiple times that I forgot the ! after the # in #![no_std] (or similar) attribute specifiers making it a useless #[no_std] causing compilation troubles because I accidentally pulled std due to this. The compiler doesn't warn for misusage of known attributes like no_std and I am not sure if it should. But it would be awesome to have a clippy lint for this situation! Would have saved me hours of debugging already - ashamed.

[Manishearth]

This is surprising, the compiler errors (or used to error?) for this kind of mistake in many cases. But yeah we should def catch any cases like this in clippy, in case we don't just make this a part of the compiler.

…

On Mon, Feb 18, 2019, 7:03 PM Hero Bird ***@***.*** wrote: When compiling a no_std crate it already happened multiple times that I forgot the ! after the # in #![no_std] (or similar) attribute specifiers causing compilation troubles because I accidentally pulled std due to this. The compiler doesn't warn for misusage of known attributes like no_std and I am not sure if it should. But it would be awesome to have a clippy lint for this situation! Would have saved me hours of debugging already - *ashamed*. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#3782>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABivSB9v0fj5Do_SsBYKDbiXisyreZKJks5vOquMgaJpZM4bA4C0> .

[phansch]

A similar lint is empty_line_after_outer_attr, but I think you had it directly above an item without any newlines, right?

[petrochenkov]

This should be an error from the compiler, yes.
We are currently lacking a centralized infra for reporting attributes applied to wrong targets though, so these errors are reported very irregularly.
I hope to implement something like rust-lang/rust#57321, but for attribute targets, some time, so this may be fixed in rustc in the future.

[J-ZhengLi]

seems like it's handled by the unused_attributes lint in rustc (which came after this issue ofc~).

So... closing it as its already done by the compiler.