Mention that -D warnings will deny ALL warnings, not just clippy warnings

[sunjay]

If we pass -D warnings to clippy, it causes the build to fail if there are any warnings, even the ones not generated by clippy. This isn't immediately obvious when you're looking at this as someone just setting up clippy, so people might not expect this nor know how to fix it. I've added a few sentences in the README to help anyone who runs into this.

These docs are useful for anyone setting up clippy warnings to be denied during CI, but still otherwise want rustc warnings to be allowed.

I could have also changed -D warnings to -D clippy::all in the Travis configurations themselves, but I wasn't sure what you would prefer to have people use as the default.

[flip1995]

We highly prefer -D warnings in travis. In Rust code you don't want to have any warnings in your crate, since these will lead to bugs.

E.g.

enum Foo {
    A,
    B,
}
//...
match some_foo {
    A => (),
    B => (),
}

will only produce a warning, telling you that B => (), is unreachable, since the first arm is a variable binding. It also suggests that you should use Foo::A instead.

dead_code warnings won't produce bugs like the one above, but if you have to many dead_code warnings in your code, you may miss warnings like the one above. Queue 2h of debugging. Been there, done that 😄

---

TL;DR: I'm willing to merge this with the typo fixed, but you should prevent any warnings from the rust compiler in production.

[sunjay]

Typo is fixed! Yes I agree about the warnings in general, but we should still warn/inform people about what we're advising them to do.

Thanks for reviewing this! :)

[flip1995]

@bors r+

[bors]

📌 Commit a7bfac7 has been approved by flip1995

[bors]

⌛ Testing commit a7bfac7 with merge bad3e08...

[bors]

☀️ Test successful - checks-travis, status-appveyor
Approved by: flip1995
Pushing bad3e08 to master...