Single branch ifs in the body of a match arm should be made explicit

[m-rph]

Single branch ifs in the body of a match arm should be made explicit

[m-rph]

Right the suggestion will introduce more errors. Could we limit it to just a warning and suspicious?

[m-rph]

I accidentally created such a bug and the only reason I caught it was because my condition was _ => if ... and not (_, 1) => if ..., and thus rustc complained about the final wildcard being unreachable.

[Alexendoo]

This may catch bugs, but it would have far too many false positives. For correct code the suggestion conflicts with match_same_arms/wildcard_in_or_patterns also

Notably rustfmt does not allow you to have PAT => if ..., it will always wrap it in a block to reduce the confusion here

[Alexendoo]

IMO the number of false positives would make it unable to be above restriction

A perhaps equivalent restriction lint would be to lint on all uses of guards in match arms

[m-rph]

But that's not the same, is it? the intention is basically to disallow pat => if and pat => { if ... {rest} } and instead recommend pat if ... => {rest}

[Alexendoo]

The other way, requiring every guard (pat if ... =>) to become pat if ... => { ... }; pat => { ... } seems equivalent but less direct than forbidding guards, so instead pat => { if ... { ... } else { ... } }

[m-rph]

That makes sense. It feels too restrictive though, doesn't it? It increases nesting by an additional level.

I think we can close this as not to be implemented, but I'd love a way to ensure this accident doesn't happen.