Indentation of if when using the return value

[moonglum]

Currently, the following code is not ok in rubocop's opinion:

x = if true
  2
else
  3
end

What Rubocop wants is the following:

x = if true
      2
    else
      3
    end

Can I configure Rubocop to take the first version instead? Because I really prefer that!

Thanks for Rubocop 👍 It helps a lot 😸

[bbatsov]

Currently not. I've said to @jonas054 last evening we should probably make this configurable, in light of a related change for case. Personally, I'm extremely fond of (which is fine by RuboCop):

x = 
  if true
    2
  else
    3
  end

Anyways, @jonas054 will you look into this?

[jonas054]

Yes.

[moonglum]

Thanks 😸

[jonas054]

No probelm. Well... some. There are two cops that report indentation related offences for the example code:

/tmp/exaple_661.rb:2:3: C: IndentationWidth: Use 2 (not -2) spaces for indentation.
  2
  ^
/tmp/exaple_661.rb:4:3: C: IndentationWidth: Use 2 (not -2) spaces for indentation.
  3
  ^
/tmp/exaple_661.rb:5:1: W: EndAlignment: end at 5, 0 is not aligned with if at 1, 4
end
^^^

The double reporting is not a big problem as long as we consider the inspected code to be incorrect, but if we want to allow it through configuration settings, then having to change two configuration parameters in two different cops becomes a problem IMO.

The way I want to solve this is to introduce a configuration parameter in the EndAlignment cop, which is a lint cop, and then just make IndentationWidth accept both styles of indentation without configuration.

[moonglum]

From a user perspective: Sounds good to me 😄

[bbatsov]

@jonas054 This sounds reasonable, but there's something more to consider - the else alignment in constructs like if/else, unless/else. Maybe there should be a separate cop for them (like the case cop).

[jonas054]

Separate cop sounds like a good idea. ElseIndentation should check that else is indented as deep as end.

[moonglum]

Thank you so much 👍