Add support for assert_true and assert_false to RSpec/Rails/MinitestAssertions

[ydah]

related to rubocop/rubocop-rspec_rails#7

This PR add support for assert_true and assert_false to RSpec/Rails/MinitestAssertions.

---

Before submitting the PR make sure the following are checked:

• Feature branch is up-to-date with master (if not - rebase it).
• Squashed related commits together.
• Added tests.
• Updated documentation.
• Added an entry to the CHANGELOG.md if the new code introduces user-observable changes.
• The build (bundle exec rake) passes (be sure to run this locally, since it may produce updated documentation that you will need to commit).

[pirj]

Looks mostly good.
I agree with @koic that we should auto-correct to stricter be(true)/be(false).

[ydah]

Ah I missed some comments. Hmm. I think we should distinguish between Minitest and Test:: Unit.
I thought it best not to support assert_true and assert_false in this Cop. It would be better to discuss the extent of this support first. For example, it looks like it could be transferred to a Cop called RSpec/TestUnitAssertions. @rubocop/rubocop-rspec WDYT?

[G-Rath]

fwiw I currently vote for "have a single cop" because these all seem to be using the same structure and as a downstream user my focus is on "use expect consistently" - that doesn't matter if the assertions come from Minitest, Test::Unit, or Rails.

Put another way: I don't think I've yet seen a technical reason for having different cops? i.e. the assertions so far have followed the same logic and structure regardless of where they're actually from.

If people want to rename it to something like RSpec/PreferExpectInsteadOfAssert, that'd be fine with me but I think having two cops based on "Minitest vs TestUnit" when they have the same behaviours is just adding more work for ourselves (and to take that a step further: should we have a third cop dedicated to Rails' extension assertions?)

[ydah]

@G-Rath Thank you for expressing your opinion.

Put another way: I don't think I've yet seen a technical reason for having different cops?

In many cases, I think there is no problem to combine them into one cop. However, as the devil's advocate, I would like to deepen this discussion.

For example, the following comments seem to be points where different actions are required. The compatibility of assert and refute is different between Minitest and Test:: Unit.

• Add support for assert_true and assert_false to RSpec/Rails/MinitestAssertions #1786 (comment)

Also, the adapter to Minitest is defined in rspec-rails, which is appropriate for the RSpec/Rails department.
https://github.com/rspec/rspec-rails/blob/main/lib/rspec/rails/adapters.rb#L140

Note, however, that Unit:: Test exists in rspec-core.
https://github.com/rspec/rspec-core/blob/f8c8880dabd8f0544a6f91d8d4c857c1bd8df903/lib/rspec/core/test_unit_assertions_adapter.rb#L9

In other words, it seems odd to enable Cop for the RSpec/Rails department if you are using Unit:: Test-style assertions and not rspec-rails.

[G-Rath]

However, as the devil's advocate, I would like to deepen this discussion.

Yup fully in support of that, and in doing so sooner rather than later.

In other words, it seems odd to enable Cop for the RSpec/Rails department if you are using Unit:: Test-style assertions and not rspec-rails

Just want to check that I'm understanding what you mean correctly, which is effectively: it's weird for someone who's using just Unit::Test but not Rails, to enable a Rails cop.

If so then yes I agree with that - I'd wondered why this was considered a Rails cop though you've pointed out it's rspec-rails that has the Minitest adapter so I'm guessing that is (or was...) the why.

[pirj]

odd to enable Cop for the RSpec/Rails department if you are using Unit:: Test-style assertions and not rspec-rails.

Fair enough, but in the future when we extract rubocop-rspec_rails, we could also extract Rails-specific assertion detectors from this cop, and make the cop open for extension.
We’ll depend on rubocop-rspec from rubocop-rspec_rails, right?

[pirj]

Thank you!

[G-Rath]

Probably best to take this to the issue, but there's also Capybara assertions we could be translating, and I'd like to also include aggregate_assertions as it seems very solid and 1:1 with aggregate_failures

[ydah]

We’ll depend on rubocop-rspec from rubocop-rspec_rails, right?

Oh, I see. If we needs to extract to rubocop-rspec_rails, maybe we shouldn't add support for this to the RSpec/Rails department's Cop for now?

[pirj]

I think this all barely exists outside of rspec-rails. Despite theexpect_with setting, rspec-rails exposes those assert methods, and there’s no known) way to hide them.

It’s ok to add all this, and all of this actually, not just the AS aliases should later be extracted to rubocop-rspec_rails.

[ydah]

I update this PR.

[pirj]

Thank you!

[bquorning]

Thank you