Prevent equality checks as statements

[jonmountjoy]

Describe the rule you'd like to see added and to what rule set
I'd like to prevent this kind of bug:

...
itemState.learningAspect == LearningAspect.learnNumber;
...

In other words, I have a statement, and that statement is an equality check, the value of which I'm discarding.

It's almost always an error (for me, it's always an error), because what I invariably want to write is:

itemState.learningAspect = LearningAspect.learnNumber;

In other words, I've used an equality check (copy pasta or finger slip or whatever) instead of an assignment.

I can't imagine a scenario when anyone would want to write that kind of statement, but I guess it's possible. In other contexts (embedded as an expression), types would usually catch the error. But as a standalone statement, the type system is quite happy. I imagine at run time it performs the equality check and discards the boolean result.

In some ways I think it's similar to unawaited_futures. I wouldn't mind an "unused_boolean" - it would cover me in this situation and force me to think more about my other methods too.

Thoughts?

(thanks!)

[eernstg]

Perhaps this should be transferred to the linter repository? It seems to be a potential enhancement of existing checks for statements with no observable effect (e.g., 25; or print;). I thought we already had some linter support for flagging such statements, but I can't see any matching lints at this time.

@pq, did we have a diagnostic for "code has no effect" (some special cases of that), but it was dropped again?

[bwilkerson]

@srawlins Who might also remember whether we used to have such a diagnostic.

[pq]

@eernstg, you may be thinking of unnecessary_statements?

[pq]

In fact, @jonmountjoy: I think unnecessary_statements covers your case?

[eernstg]

Exactly, unnecessary_statements is what I was looking for, I just didn't come up with the right words to search for. ;-)

[jonmountjoy]

Ahhh that's fabulous @eernstg @pq @bwilkerson - that does the job!
I would seriously consider putting this in the default recommended lints.

This is so nice - thank you!