Warn if classifier instance is out of date

lib/linguist/classifier.rb

@@ -29,6 +29,19 @@ module Linguist
       @languages       = Hash.new(0)
     end
 
+    # Public: Compare Classifier objects.
+    #
+    # other - Classifier object to compare to.
+    #
+    # Returns Boolean.
+    def eql?(other)
+      # Lazy fast check counts only
+      other.is_a?(self.class) &&
+        @tokens_total == other.instance_variable_get(:@tokens_total) &&
+        @languages_total == other.instance_variable_get(:@languages_total)
+    end
+    alias_method :==, :eql?
+
     # Public: Train classifier that data is a certain language.
     #
     # language - Language of data

test/test_classifier.rb

@@ -16,6 +16,13 @@ class TestClassifier < Test::Unit::TestCase
     File.read(File.join(fixtures_path, name))
   end
 
+  def test_instance_freshness
+    # Just warn, it shouldn't scare people off by breaking the build.
+    unless Classifier.instance.eql?(Linguist::Sample.classifier)
+      warn "Classifier database is out of date. Run `bundle exec rake classifier`."
+    end
+  end
+
   def test_classify
     classifier = Classifier.new
     classifier.train Language["Ruby"], fixture("ruby/foo.rb")