Improve performance for many unresolved includes/extends

Potential fix for ruby#1126

lib/rdoc/mixin.rb

@@ -18,6 +18,7 @@ def initialize(name, comment)
     @name = name
     self.comment = comment
     @module = nil # cache for module if found
+    @module_lookup_failed = false
   end
 
   ##
@@ -64,16 +65,10 @@ def inspect # :nodoc:
   # - if not found, look into the children of included modules,
   #   in reverse inclusion order;
   # - if still not found, go up the hierarchy of names.
-  #
-  # This method has <code>O(n!)</code> behavior when the module calling
-  # include is referencing nonexistent modules.  Avoid calling #module until
-  # after all the files are parsed.  This behavior is due to ruby's constant
-  # lookup behavior.
-  #
-  # As of the beginning of October, 2011, no gem includes nonexistent modules.
 
   def module
     return @module if @module
+    return @name if @module_lookup_failed
 
     # search the current context
     return @name unless parent
@@ -101,6 +96,7 @@ def module
       up = up.parent
     end
 
+    @module_lookup_failed = true
     @name
   end
 

test/rdoc/test_rdoc_context.rb

@@ -409,13 +409,26 @@ def test_add_to_done_documenting
     refute_includes arr, incl
   end
 
-  def bench_add_include
-    cm = RDoc::ClassModule.new 'Klass'
+  def test_bench_add_include
+    assert_linear_performance (1..25) do |count|
+      cm = RDoc::ClassModule.new 'Klass'
+      count.times do |i|
+        cm.add_include RDoc::Include.new("N::M#{i}", nil)
+      end
+    end
+  end
+
+  def test_bench_include_module_resolution
+    assert_linear_performance (1..25) do |count|
+      cm = RDoc::ClassModule.new 'Klass'
+      cm.store = RDoc::Store.new
 
-    assert_performance_linear 0.5 do |count|
       count.times do |i|
         cm.add_include RDoc::Include.new("N::M#{i}", nil)
       end
+      last_include = RDoc::Include.new("N::M#{count + 1}", nil)
+      cm.add_include(last_include)
+      last_include.module
     end
   end