Wanted distinct column values via a has_many relation. Needed #select and #pluck.

student.rb

class Student < ActiveRecord::Base
  belongs_to :teacher

  # neighborhood: string
end

teacher.rb

class Teacher < ActiveRecord::Base
  has_many :students

  def neighborhoods
    self.students.select(:neighborhood).uniq.pluck(:neighborhood)
  end
end

Adding the select(:neighborhood) allowed the uniq to generate the DISTINCT in the SQL query.

Reordering it to:

self.students.pluck(:neighborhood).uniq

would have been OK, but the uniq would have been performed in Ruby on an array. Probably OK, but might as well use the db to do this.

Oh, and if the args to select and pluck are the same string, then the query will collapse the fields.

t = Teacher.first
t.students.select(:neighborhood).uniq.pluck(:neighborhood)

yields

SELECT DISTINCT neighborhood, "students"."neighborhood" FROM "students"...

and

t = Teacher.first
t.students.select('neighborhood').uniq.pluck('neighborhood')

yields

SELECT DISTINCT neighborhood FROM "students"...

both return the same result, but the latter is the query we'd actually want.