False negatives for `Rake/Desc` cop

[ydakuka]

Actual behavior

task lint: %i[rubocop]

terminal:

ydakuka@yauhenid:~/Work/project$ bin/rails_docker rubocop Rakefile 
Inspecting 1 file
.

1 file inspected, no offenses detected

Rubocop

ydakuka@yauhenid:~/Work/main_app$ bin/rails_docker rubocop -V
1.57.1 (using Parser 3.2.2.4, rubocop-ast 1.29.0, running on ruby 2.7.8) [x86_64-linux]
  - rubocop-capybara 2.19.0
  - rubocop-factory_bot 2.24.0
  - rubocop-performance 1.19.1
  - rubocop-rails 2.21.1
  - rubocop-rake 0.6.0
  - rubocop-rspec 2.24.1
  - rubocop-thread_safety 0.5.1

[pboling]

What is the expected behavior? The problem is unclear.

Is this the same as #46 ?

[ydakuka]

I have the following code:

task lint: :rubocop

I run rubocop and get offences:

ydakuka@yauhenid:~/Work/project$ bin/rails_docker rubocop Rakefile
Inspecting 1 file
C

Offenses:

Rakefile:3:1: C: Rake/Desc: Describe the task with desc method.
task lint: :rubocop
^^^^^^^^^^^^^^^^^^^

1 file inspected, 1 offense detected

However, if I have the code:

task lint: %i[rubocop]

I will run rubocop and will not receive any offences.

ydakuka@yauhenid:~/Work/project$ bin/rails_docker rubocop Rakefile
Inspecting 1 file
.

1 file inspected, no offenses detected

[mvz]

I'm trying to understand the desired behavior here, and I'm puzzled by the idea that having one prerequisite should trigger this offense, but multiple prerequisites are fine.

[pboling]

@mvz I think the issue is in the determination of prerequisites, which are exempted from needing a desc, but only if they are set as an Array. See: https://github.com/rubocop/rubocop-rake/blob/master/lib/rubocop/cop/rake/desc.rb#L36-L48

        def_node_matcher :prerequisites, <<~PATTERN
          (send nil? :task (hash (pair _ $_)))
        PATTERN

        def on_task(node)
          return if task_with_desc?(node)
          return if Helper::TaskName.task_name(node) == :default

          requirements = prerequisites(node)
          return if requirements&.array_type?

          add_offense(node)
        end

[mvz]

@pboling indeed, but what's the reasoning behind this difference?

[pboling]

I expect it is perhaps due to a logical trap. Perhaps they couldn't think of how to identify the non-array version without picking up other unwanted matches that would break other tests or cops. I expect the thinking was having a little is better than nothing...