Request for automatic invocation of 'reset_column_information' in Migrations

I concur. Just ran into this myself.

Spent a few hours hunting it down and another hour making updates to my migrations.

In the rails test suite, there is a test where they specifically call reset_column_information:

def test_native_types
      Person.delete_all
      Person.connection.add_column "people", "last_name", :string
      Person.connection.add_column "people", "bio", :text
      Person.connection.add_column "people", "age", :integer
      Person.connection.add_column "people", "height", :float
      Person.connection.add_column "people", "wealth", :decimal, :precision => '30', :scale => '10'
      Person.connection.add_column "people", "birthday", :datetime
      Person.connection.add_column "people", "favorite_day", :date
      Person.connection.add_column "people", "moment_of_truth", :datetime
      Person.connection.add_column "people", "male", :boolean
      Person.reset_column_information

      assert_nothing_raised do
        Person.create :first_name => 'bob', :last_name => 'bobsen',
          :bio => "I was born ....", :age => 18, :height => 1.78,
          :wealth => BigDecimal.new("12345678901234567890.0123456789"),
          :birthday => 18.years.ago, :favorite_day => 10.days.ago,
          :moment_of_truth => "1782-10-10 21:40:18", :male => true
      end

      # Other stuff
end

I think this is just the way it's supposed to be. One reason could be that otherwise they'd have to call reset_column_information after every single operation (add_column, remove_column, etc).

Well, this feature request is for running the method after every Migration; not within the migration after each Model modification. But even then:

irb(main):008:0> puts Benchmark.measure {1000000.times {User.reset_column_information;}}
  3.050000   0.010000   3.060000 (  3.079835)

I'm not worried about the performance impact of this, the more so since we are dealing on a relatively infrequent operation as running migrations.