Linter for length(x == y) where almost certainly length(which(x == y)) or length(x) == y was intended

[MichaelChirico]

This is a mistake that I find myself making all the time:

if (length(x == 0)) ...

It makes itself into a really subtle bug and many wasted minutes.

I can't think of a case where length(x == y) is preferable to length(x) or length(y), unless there's some weird coercion going on and that should be made explicit anyway.

Therefore it seems like a good candidate for a new linter.

I don't really know what to name such a linter... I also wonder if maybe it's good for an umbrella linter like common_mistakes_linter(), though I can't really think of other good candidate rules for that off the top of my head. see comment below

[AshesITR]

How about length_equals_linter()?
I imagine the most common intention is length(x) == y.

I don't think we should make an umbrella linter.

[MichaelChirico]

Note to self: the right way to group this is with the tag common_mistakes.

[AshesITR]

BTW the length(which()) case should be replaced by sum().

[MichaelChirico]

BTW the length(which()) case should be replaced by sum().

I think that's a separate linter... there's some nuance there, as length(which(A)) is equivalent to sum(A) but the most common case will be comparing length(which(A)) to 0, in which case any(A) is preferable, which is boolean_arithmetic_linter().