New cop for .. operator precedence

[pocke]

The following code has a bug.

# test.rb
def foo(x = nil)
  x || 1..2
end

p foo
p foo(1..3)

$ ruby test.rb
1..2
test.rb:3:in `foo': bad value for range (ArgumentError)
	from test.rb:7:in `<main>'

The code is interpreted as (x || 1)..2, but it looks like x || (1..2) to me.
I think it is confusing, so, we need a cop for this problem. However we don't have the cop.

The cause is .. operator's precedence. https://ruby-doc.org/core-2.2.0/doc/syntax/precedence_rdoc.html

---

Expected behavior

x || 1..2

RuboCop adds an offence for the above code.

Actual behavior

The cop for this problem does not exist.

RuboCop version

$ rubocop -V
0.48.0 (using Parser 2.4.0.0, running on ruby 2.4.1 x86_64-linux)

[Drenmi]

It can also be unexpected on the "other side" of the range:

1..2.to_a

will attempt to send #to_a to 2, rather than the range.

On that note, I think the cop should be a parenthesizing cop (possibly with an additional ambiguity cop.)

I've seen some code like this in RuboCop:

(foo.bar)..(baz.qux)

and we don't have any cops for enforcing a style. I also don't think there is a Style Guide entry. 🙂

[bbatsov]

True. I've made this mistake like a million times. :-)

…

On Sun, Apr 2, 2017 at 08:53 Ted Johansson ***@***.***> wrote: It can also be unexpected on the "other side" of the range: 1..2.to_a will attempt to send #to_a to 2, rather than the range. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4182 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAGVypdJu6kYYnTQvP_UcpfGEeev0lfAks5rrzfcgaJpZM4MqOjD> .

[denys281]

@Drenmi is it actual issue?