Manual assert_eq! (enhance `if_then_panic`)

[camsteffen]

We just added if_then_panic, but then I realized there's a possible enhancement for assert_eq!/assert_ne!:

// before
if a != b {
    panic!("hi");
}
// after
assert_eq!(a, b, "hi");

I think a rename to manual_assert would be best. But we could also keep the current name (it still kinda makes sense) or create a new lint for this case.

[xFrednet]

Nice suggestion, and renaming the lint to manual_assert sounds like a good idea. We should try to do this, before the if_then_panic lint hit's stable, IMO 🙃

[Labelray]

I am interested in implementing the enhancement. And I think it is better to use just one lint manual_assert for current if_then_panic and this new case, not new lint for this new case.

@rustbot claim

[xFrednet]

Hey @Labelray, if you're working on this issue, you might also want to take a look at #7718. @mikerite suggested in the issue to move this lint to pedantic which seems reasonable to me. That might be an easy change to do along the way 🙃.

[camsteffen]

Another thing I thought of - lint if-else-panic and don't lint if-panic-else-panic.

[jplatte]

Relatedly, if !something { panic!(); } should suggest assert!(something); instead of assert!(!!something);. (saw this in https://github.com/ruma/ruma/runs/3759040805?check_suite_focus=true#step:5:399, I'm guessing that the double negation would be happening without the binop which is also not being treated correctly (#7731).

[spearman]

An issue it seems with the current manual_assert is that it suggests using assert for types that are not Debug.

[abendrothj]

Hi, is this still needed? I'd be happy to take a look.