incomplete Marshal.dump

an evil ivar_as_index is added to Object to hold
the names of modules that extend the object

kernel/bootstrap/object.rb

@@ -167,4 +167,8 @@ def freeze
   def frozen?
     Ruby.primitive :object_frozen_p
   end
+
+  def add_module_name(mod_name)
+    Ruby.primitive :add_module_name
+  end
 end

kernel/core/marshal.rb

@@ -5,5 +5,157 @@ module Marshal
   MAJOR_VERSION = 4
   MINOR_VERSION = 8
 
-end
+  TYPE_NIL = '0'
+  TYPE_TRUE = 'T'
+  TYPE_FALSE = 'F'
+  TYPE_FIXNUM = 'i'
+
+  TYPE_EXTENDED = 'e'
+  TYPE_UCLASS = 'C'
+  TYPE_OBJECT = 'o'
+  TYPE_DATA = 'd'  # no specs
+  TYPE_USERDEF = 'u'
+  TYPE_USRMARSHAL = 'U'  # no specs
+  TYPE_FLOAT = 'f'
+  TYPE_BIGNUM = 'l'
+  TYPE_STRING = '"'
+  TYPE_REGEXP = '/'
+  TYPE_ARRAY = '['
+  TYPE_HASH = '{'
+  TYPE_HASH_DEF = '}'
+  TYPE_STRUCT = 'S'
+  TYPE_MODULE_OLD = 'M'  # no specs
+  TYPE_CLASS = 'c'
+  TYPE_MODULE = 'm'
+
+  TYPE_SYMBOL = ':'
+  TYPE_SYMLINK = ';'
+
+  TYPE_IVAR = 'I'
+  TYPE_LINK = '@'
+
+  def self.dump(obj, depth = -1, io = nil)
+    str = [MAJOR_VERSION, MINOR_VERSION].pack('CC')
+    str << serialize(obj, depth)
+    str
+  end
+
+  def self.serialize(obj, depth)
+    get_class_names(obj).each do |class_name|
+      case class_name
+      when 'NilClass'
+        return serialize_nil(obj)
+      when 'TrueClass'
+        return serialize_true(obj)
+      when 'FalseClass'
+        return serialize_false(obj)
+      when 'Class'
+        return serialize_class(obj)
+      when 'Module'
+        return serialize_module(obj)
+      when 'Symbol'
+        return serialize_symbol(obj)
+      when 'String'
+        return serialize_string(obj)
+      when / Integer | Fixnum | Bignum /x
+        if obj >= -2**30 and obj <= (2**30 - 1)
+          return serialize_fixnum(obj)
+        end
+      end
+    end
+    ''
+  end
+
+  def self.serialize_integer(n)
+    if n == 0
+      s = to_byte(n)
+    elsif n > 0 and n < 123
+      s = to_byte(n + 5)
+    elsif n < 0 and n > -124
+      s = to_byte(256 + (n - 5))
+    elsif n < 0
+      s = "\x00"
+    else
+      s = "\x00"
+      cnt = 0
+      while n != 0
+        s << to_byte(n & 0xFF)
+        n >>= 8
+        cnt += 1
+      end
+      s[0] = to_byte(cnt)
+    end
+    s
+  end
+
+  def self.serialize_nil(obj)
+    TYPE_NIL
+  end
+
+  def self.serialize_true(obj)
+    TYPE_TRUE
+  end
+
+  def self.serialize_false(obj)
+    TYPE_FALSE
+  end
 
+  def self.serialize_symbol(obj)
+    str = obj.to_s
+    TYPE_SYMBOL + serialize_integer(str.length) + str
+  end
+
+  def self.serialize_class(obj)
+    TYPE_CLASS + serialize_integer(obj.name.length) + obj.name
+  end
+
+  def self.serialize_module(obj)
+    TYPE_MODULE + serialize_integer(obj.name.length) + obj.name
+  end
+
+  def self.serialize_extended_object(obj)
+    str = ''
+    if obj.module_names.kind_of?(Array)
+      obj.module_names.each do |mod_name|
+        str << TYPE_EXTENDED + serialize_symbol(mod_name)
+      end
+    end
+    str
+  end
+
+  def self.serialize_user_class(obj, cls)
+    if obj.class != cls
+      TYPE_UCLASS + serialize_symbol(obj.class.name)
+    else
+      ''
+    end
+  end
+
+  def self.serialize_string(obj)
+    serialize_extended_object(obj) +
+    serialize_user_class(obj, String) +
+    TYPE_STRING + serialize_integer(obj.length) + obj
+  end
+
+  def self.serialize_fixnum(obj)
+    TYPE_FIXNUM + serialize_integer(obj)
+  end
+
+  def self.get_class_names(obj)
+    cls = obj.class
+    sup = get_superclass(cls)
+    [cls.to_s, sup.to_s]
+  end
+
+  def self.get_superclass(cls)
+    sup = cls.superclass
+    while sup and sup.superclass and sup.superclass != Object
+      sup = sup.superclass
+    end
+    sup
+  end
+
+  def self.to_byte(n)
+    [n].pack('C')
+  end
+end

kernel/core/object.rb

@@ -8,8 +8,9 @@ class Object
   RUBY_ENGINE = Rubinius::RUBY_ENGINE
   RBX_VERSION = Rubinius::RBX_VERSION
 
-  ivar_as_index :__ivars__ => 0
+  ivar_as_index :__ivars__ => 0, :module_names => 1
   def __ivars__; @__ivars__ ; end
+  def module_names; @module_names; end
 
   def initialize
   end
@@ -249,6 +250,7 @@ def protected_methods(all=true)
   def extend(*modules)
     modules.reverse_each do |mod|
       mod.extend_object(self)
+      add_module_name(mod.name)
     end
     self
   end

lib/compiler/system_hints.rb

@@ -7,7 +7,7 @@ module Bootstrap
     :MetaClass=>{:@__ivars__=>0, :@methods=>1, :@method_cache=>2, :@name=>3, :@constants=>4, :@parent=>5, :@superclass=>6, :@instance_fields=>7, :@has_ivars=>8, :@needs_cleanup=>9, :@object_type=>10, :@attached_instance=>11},
     :Class=>{:@__ivars__=>0, :@methods=>1, :@method_cache=>2, :@name=>3, :@constants=>4, :@parent=>5, :@superclass=>6, :@instance_fields=>7, :@has_ivars=>8, :@needs_cleanup=>9, :@object_type=>10},
     :BlockContext=>{},
-    :Object=>{:@__ivars__=>0},
+    :Object=>{:@__ivars__=>0, :@module_names => 1},
     :Float=>{:@__ivars__=>0},
     :Array=>{:@total=>0, :@tuple=>1, :@start => 2, :@shared => 3},
     :String=>{:@bytes=>0, :@characters=>1, :@encoding=>2, :@data=>3, :@hash => 4, :@shared => 5},

shotgun/lib/primitive_names.rb

@@ -180,7 +180,8 @@ class Compiler
       :bignum_from_float,
       :task_stack_size,
       :task_get_stack_value,
-      :fastctx_reload_method
+      :fastctx_reload_method,
+      :add_module_name
     ]
 
     FirstRuntimePrimitive = 1024

shotgun/lib/primitives.rb

@@ -2898,6 +2898,22 @@ def object_send
     CODE
   end
 
+  def add_module_name
+    <<-CODE
+    self = stack_pop();
+    t1 = stack_pop();
+    if(!RISA(self, class) && !RISA(self, module)) {
+      t2 = object_get_module_names(self);
+      if(!RISA(t2, array)) {
+        t2 = array_new(state, 1);
+        object_set_module_names(self, t2);
+      }
+      array_append(state, t2, t1);
+    }
+    stack_push(self);
+    CODE
+  end
+
 end
 
 prim = ShotgunPrimitives.new

spec/data/ruby/1.8/core/marshal/dump_excludes.txt

@@ -2,42 +2,27 @@ Marshal.dump when given a recursion limit raises an ArgumentError when the recur
 Marshal.dump when given an IO-Object writes the serialized data to the IO-Object
 Marshal.dump when given an IO-Object returns the IO-Object
 Marshal.dump when given an IO-Object raises an Error when the IO-Object does not respond to #write
-Marshal.dump returns a string-serialized version of the given argument
 Marshal.dump raises an Error when trying to dump an anonymous class/module
-Marshal.dump with nil returns a string-serialized version of the given argument
-Marshal.dump with true returns a string-serialized version of the given argument
-Marshal.dump with false returns a string-serialized version of the given argument
-Marshal.dump with empty string returns a string-serialized version of the given argument
-Marshal.dump with string of length 100 returns a string-serialized version of the given argument
-Marshal.dump with string of length 200 returns a string-serialized version of the given argument
-Marshal.dump with string of length 300 returns a string-serialized version of the given argument
-Marshal.dump with empty extended_string returns a string-serialized version of the given argument
-Marshal.dump with empty user_string returns a string-serialized version of the given argument
-Marshal.dump with empty extended_user_string returns a string-serialized version of the given argument
-Marshal.dump with symbol of length 1 returns a string-serialized version of the given argument
-Marshal.dump with symbol of length 300 returns a string-serialized version of the given argument
-Marshal.dump with fixnum 0 returns a string-serialized version of the given argument
-Marshal.dump with fixnum 5 returns a string-serialized version of the given argument
-Marshal.dump with fixnum 2**16 returns a string-serialized version of the given argument
-Marshal.dump with fixnum 2**24 returns a string-serialized version of the given argument
-Marshal.dump with fixnum (2**30 - 1) returns a string-serialized version of the given argument
+Marshal.dump with fixnum -2**8 returns a string-serialized version of the given argument
+Marshal.dump with fixnum -16711680 returns a string-serialized version of the given argument
 Marshal.dump with fixnum -2**16 returns a string-serialized version of the given argument
+Marshal.dump with fixnum -(2**16 - 1) returns a string-serialized version of the given argument
+Marshal.dump with fixnum -(2**16 - 2) returns a string-serialized version of the given argument
+Marshal.dump with fixnum -43690 returns a string-serialized version of the given argument
 Marshal.dump with fixnum -2**30 returns a string-serialized version of the given argument
 Marshal.dump with bignum 2**30 returns a string-serialized version of the given argument
-Marshal.dump with bignum -2**31 returns a string-serialized version of the given argument
+Marshal.dump with bignum -(2**31 -1) returns a string-serialized version of the given argument
 Marshal.dump with bignum -2**63 returns a string-serialized version of the given argument
 Marshal.dump with bignum 2**64 returns a string-serialized version of the given argument
 Marshal.dump with bignum -2**64 returns a string-serialized version of the given argument
-Marshal.dump with class returns a string-serialized version of the given argument
-Marshal.dump with module returns a string-serialized version of the given argument
 Marshal.dump with any object having _dump method returns a string-serialized version of the given argument
 Marshal.dump with any extended_object having _dump method returns a string-serialized version of the given argument
 Marshal.dump with object returns a string-serialized version of the given argument
 Marshal.dump with object having ivar returns a string-serialized version of the given argument
 Marshal.dump with regexp returns a string-serialized version of the given argument
 Marshal.dump with extended_regexp returns a string-serialized version of the given argument
 Marshal.dump with user_regexp having option Regexp::IGNORECASE returns a string-serialized version of the given argument
-Marshal.dump with extended_user_regexp returns a string-serialized version of the given argument
+Marshal.dump with extended_user_regexp having ivar returns a string-serialized version of the given argument
 Marshal.dump with float 0.0 returns a string-serialized version of the given argument
 Marshal.dump with float NaN returns a string-serialized version of the given argument
 Marshal.dump with float -Infinity returns a string-serialized version of the given argument
@@ -50,7 +35,7 @@ Marshal.dump with extended_user_hash_default returns a string-serialized version
 Marshal.dump with array returns a string-serialized version of the given argument
 Marshal.dump with array containing the same symbols returns a string-serialized version of the given argument
 Marshal.dump with array containing refs to the same object returns a string-serialized version of the given argument
-Marshal.dump with extended_array returns a string-serialized version of the given argument
+Marshal.dump with extended_array having ivar returns a string-serialized version of the given argument
 Marshal.dump with user_array returns a string-serialized version of the given argument
 Marshal.dump with extended_user_array returns a string-serialized version of the given argument
 Marshal.dump with struct returns a string-serialized version of the given argument