Discuss and setting up Style/HashAsLastArrayItem cop

[roman-dubrovsky]

Is your feature request related to a problem? Please describe.
Here I just want to discuss rule for this cop.

This cop sets up a rule for hash braces in the array.

EnforcedStyle: braces (default)

# bad
[1, 2, one: 1, two: 2]

# good
[1, 2, { one: 1, two: 2 }]

EnforcedStyle: no_braces (our rule)

# bad
[1, 2, { one: 1, two: 2 }]

# good
[1, 2, one: 1, two: 2]

cons of each rule on real examples:

For braces rule

In the include blocks we need to add hash braces

Post.unscoped.includes({account_holders: [{user: :profile}]})

Usually, I write something like this

Post.unscoped.includes(account_holders: [user: :profile])

For no_braces rule, usually, I faced the issue on creating an expected response for any API

Instead of writing some array which contains list of object

let(:contacts_list) do
  {
    contacts: [{contactID: 'test'}]
  }
end

I need to write this variant

let(:contacts_list) do
  {
    contacts: [contactID: 'test']
  }
end

But if says a truth it can be really easy refactored

let(:contact) { {contactID: 'test'}  }
let(:contacts_list) { {contacts: [contact] } }

What do you think about it?

[roman-dubrovsky]

P.S. Also right now no_braces has a bug for array of hashes. It converts this expression

contacts = [{contactID: 'test'}, {contactID: 'test2'}]

to this one

contacts = [contactID: 'test', contactID: 'test2']

which is equal [{:contactID=>"test2"}]

But I'm going to report this issue to rubocop library

P.P.S. Seems like this issue has been fixed rubocop/rubocop#8323

[roman-dubrovsky]

Personally, I still love no_braces rule more

[texpert]

I'm also for no_braces.

[DDKatch]

[1, 2, one: 1, two: 2] may mislead to think that the real value here is [1, 2, {one: 1}, {two: 2}] but actually [1, 2, {one: 1, two: 2}]. It may lead to mistakes.

[DDKatch]

[contactID: 'test'] convertes to [{:contactID=>"test2"}]. I always thought it's how ruby works 🤔
The issue(rubocop/rubocop#8323) was about parsing errors for a multiline array of hashes

[roman-dubrovsky]

[1, 2, one: 1, two: 2] may mislead to think that the real value here is [1, 2, {one: 1}, {two: 2}] but actually [1, 2, {one: 1, two: 2}]. It may lead to mistakes.

@DDKatch are you sure? in you case, you should write something like this (by this cop) [1, 2, {one: 1}, two: 2]. It looks suck but it's easy for refactoring or we can patch this rubocop rule in the future (e.g. restrict this rule for multiple array)

[contactID: 'test'] convertes to [{:contactID=>"test2"}]. I always thought it's how ruby works 🤔
The issue(rubocop/rubocop#8323) was about parsing errors for a multiline array of hashes

in my example above, ruby just overrides the value for contactID key . Or I didn't understand your issue 🤔

[roman-dubrovsky]

this issue has been fixed #199 (comment)

from documentation:

# good
[{ one: 1 }, { two: 2 }]

So I'm going to close this issue and leave no_braces style