Internal encoding can only be UTF-8.

core/encoding.rb

@@ -517,15 +517,11 @@ def self.default_external=(enc)
   end
 
   def self.default_internal
-    if undefined.equal? @default_internal
-      @default_internal = find "internal"
-    end
-    @default_internal
+    # Rubinius internal encoding is always UTF-8
   end
 
   def self.default_internal=(enc)
-    set_alias_index "internal", enc
-    @default_internal = undefined
+    Rubinius::Logger.system.warn "Encoding.default_internal= is deprecated. Rubinius internal encoding is always UTF-8"
   end
 
   def self.find(name)

core/loader.rb

@@ -296,15 +296,27 @@ def options(argv=ARGV)
         @load_paths << dir
       end
 
-      options.on "-K", "Ignored $KCODE option for compatibility"
+      options.on "-K", "Ignored $KCODE option for compatibility" do
+        msg = "The -K option is deprecated. Rubinius internal encoding is always UTF-8"
+        STDERR.puts msg
+        Rubinius::Logger.system.warn msg
+      end
+
       options.on "-U", "Set Encoding.default_internal to UTF-8" do
-        set_default_internal_encoding('UTF-8')
+        msg = "The -U option is deprecated. Rubinius internal encoding is always UTF-8"
+        STDERR.puts msg
+        Rubinius::Logger.system.warn msg
       end
 
       options.on "-E", "ENC", "Set external:internal character encoding to ENC" do |enc|
-        ext, int = enc.split(":")
-        Encoding.default_external = ext if ext and !ext.empty?
-        set_default_internal_encoding(int) if int and !int.empty?
+        external, internal = enc.split(":")
+        Encoding.default_external = external if external and !external.empty?
+
+        if internal and !internal.empty?
+          msg = "The -E option setting internal encoding is deprecated. Rubinius internal encoding is always UTF-8"
+          STDERR.puts msg
+          Rubinius::Logger.system.warn msg
+        end
       end
 
       options.on "--main", "PATH", "Load PATH directly from CodeDB" do |path|
@@ -495,16 +507,6 @@ def set_program_name(name)
     end
     private :set_program_name
 
-    def set_default_internal_encoding(encoding)
-      if @default_internal_encoding_set && Encoding.default_internal.name != encoding
-        raise RuntimeError, "Default internal encoding already set to '#{Encoding.default_internal.name}'."
-      else
-        @default_internal_encoding_set = true
-        Encoding.default_internal = encoding
-      end
-    end
-    private :set_default_internal_encoding
-
     def handle_rubyopt(options)
       if env_opts = ENV['RUBYOPT']
         Rubinius::Logger.system.write "RUBYOPT: #{env_opts}"

core/string.rb

@@ -1102,6 +1102,10 @@ def encode!(to=undefined, from=undefined, options=undefined)
       options = Rubinius::Type.coerce_to options, Hash, :to_hash
     end
 
+    # TODO: Only UTF-8 is allowed for internal encoding. This is a preliminary
+    # step to fixing all encoding-related interfaces.
+    to_enc = Encoding::UTF_8 unless to_enc == Encoding::UTF_8
+
     if ascii_only? and from_enc.ascii_compatible? and to_enc and to_enc.ascii_compatible?
       force_encoding to_enc
     elsif to_enc and from_enc != to_enc
@@ -1152,6 +1156,10 @@ def end_with?(*suffixes)
 
   def force_encoding(enc)
     @encoding = Rubinius::Type.coerce_to_encoding enc
+
+    # TODO: Only UTF-8 encodings are supported internally.
+    return self unless @encoding.equal?(Encoding::UTF_8) or @encoding.equal?(Encoding::BINARY)
+
     unless @ascii_only && @encoding.ascii_compatible?
       @ascii_only = @valid_encoding = @num_chars = nil
     end
@@ -1160,6 +1168,7 @@ def force_encoding(enc)
       @valid_encoding = true
       @num_chars = 0
     end
+
     self
   end
 

machine/capi/encoding.cpp

@@ -99,13 +99,8 @@ extern "C" {
   }
 
   rb_encoding *rb_default_internal_encoding(void) {
-    NativeMethodEnvironment* env = NativeMethodEnvironment::get();
-    Encoding* enc = Encoding::find(env->state(), "internal");
-    if(enc->nil_p()) {
-      return 0;
-    } else {
-      return enc->encoding();
-    }
+    // Rubinius internal encoding is always UTF-8
+    return 0;
   }
 
   rb_encoding *rb_default_external_encoding(void) {

machine/class/encoding.cpp

@@ -58,7 +58,6 @@ namespace rubinius {
     G(encoding)->set_const(state, "EncodingList", G(encoding_list));
 
     G(encoding)->set_ivar(state, state->symbol("@default_external"), G(undefined));
-    G(encoding)->set_ivar(state, state->symbol("@default_internal"), G(undefined));
     G(encoding)->set_ivar(state, state->symbol("@filesystem_encoding"), G(undefined));
 
     Encoding* binary = create_bootstrap(state, "ASCII-8BIT", eBinary, ONIG_ENCODING_ASCII);
@@ -229,16 +228,8 @@ namespace rubinius {
   }
 
   Encoding* Encoding::default_internal(STATE) {
-    Encoding* enc;
-    Symbol* default_internal = state->symbol("default_internal");
-    Object* obj = G(encoding)->get_ivar(state, default_internal);
-
-    if(!(enc = try_as<Encoding>(obj))) {
-      enc = Encoding::find(state, "internal");
-      G(encoding)->set_ivar(state, default_internal, enc);
-    }
-
-    return enc;
+    // Rubinius internal encoding is always UTF-8
+    return nil<Encoding>();
   }
 
   Encoding* Encoding::filesystem_encoding(STATE) {

spec/core/mirror/string/byte_index_spec.rb

@@ -55,12 +55,5 @@
     it "returns nil if the search String is not a complete character" do
       string_mirror("あそこ").byte_index("\xe3\x82").should be_nil
     end
-
-    it "raises an ArgumentError if the encoding of String and pattern are incompatible" do
-      pattern = "こ".encode Encoding::EUC_JP
-      lambda do
-        string_mirror("あそこ").byte_index pattern
-      end.should raise_error(ArgumentError)
-    end
   end
 end

spec/core/string/byte_to_character_index_spec.rb

@@ -6,20 +6,14 @@
 
     before :each do
       @utf8 = Rubinius::Mirror.reflect("hello 椎名深夏" * 1000)
-      @utf32 = Rubinius::Mirror.reflect(@utf8.object.encode(Encoding::UTF_32BE))
-      @shift_jis = Rubinius::Mirror.reflect(@utf8.object.encode(Encoding::SHIFT_JIS))
     end
 
     it "returnse character index for the character containing the byte at the given byte index" do
       @utf8.byte_to_character_index(358).should == 200
-      @utf32.byte_to_character_index(358).should == 89
-      @shift_jis.byte_to_character_index(358).should == 257
     end
 
     it "returns the character index offset for the character containing the byte at the given byte index searching from given byte index" do
       @utf8.byte_to_character_index(234, 642).should == 130
-      @utf32.byte_to_character_index(234, 1432).should == 58
-      @shift_jis.byte_to_character_index(234, 500).should == 168
     end
 
   end

spec/core/string/character_to_byte_index_spec.rb

@@ -6,20 +6,14 @@
 
     before :each do
       @utf8 = Rubinius::Mirror.reflect("hello 椎名深夏" * 1000)
-      @utf32 = Rubinius::Mirror.reflect(@utf8.object.encode(Encoding::UTF_32BE))
-      @shift_jis = Rubinius::Mirror.reflect(@utf8.object.encode(Encoding::SHIFT_JIS))
     end
 
     it "returns the byte index for the given character index" do
       @utf8.character_to_byte_index(358).should == 642
-      @utf32.character_to_byte_index(358).should == 1432
-      @shift_jis.character_to_byte_index(358).should == 500
     end
 
     it "returns the byte index for the given character index searching from given byte index" do
       @utf8.character_to_byte_index(234, 642).should == 1064
-      @utf32.character_to_byte_index(234, 1432).should == 2368
-      @shift_jis.character_to_byte_index(234, 500).should == 828
     end
 
   end

spec/ruby/command_line/dash_upper_e_spec.rb

@@ -1,7 +1,6 @@
 describe "ruby -E" do
   it "raises a RuntimeError if used with -U" do
-    ruby_exe("p 1",
-             :options => '-Eascii:ascii -U',
-             :args => '2>&1').should =~ /RuntimeError/
+    ruby_exe('', :options => '-Eascii:ascii -U', :args => '2>&1').should == \
+      "The -E option setting internal encoding is deprecated. Rubinius internal encoding is always UTF-8\nThe -U option is deprecated. Rubinius internal encoding is always UTF-8\n"
   end
 end

spec/ruby/command_line/dash_upper_k_spec.rb

@@ -1,33 +1,6 @@
 describe 'The -K command line option sets __ENCODING__' do
-  it "to Encoding::ASCII_8BIT with -Ka" do
-    ruby_exe("print __ENCODING__", :options => '-Ka').should == Encoding::ASCII_8BIT.to_s
-  end
-
-  it "to Encoding::ASCII_8BIT with -KA" do
-    ruby_exe("print __ENCODING__", :options => '-KA').should == Encoding::ASCII_8BIT.to_s
-  end
-
-  it "to Encoding::EUC_JP with -Ke" do
-    ruby_exe("print __ENCODING__", :options => '-Ke').should == Encoding::EUC_JP.to_s
-  end
-
-  it "to Encoding::EUC_JP with -KE" do
-    ruby_exe("print __ENCODING__", :options => '-KE').should == Encoding::EUC_JP.to_s
-  end
-
-  it "to Encoding::UTF_8 with -Ku" do
-    ruby_exe("print __ENCODING__", :options => '-Ku').should == Encoding::UTF_8.to_s
-  end
-
-  it "to Encoding::UTF_8 with -KU" do
-    ruby_exe("print __ENCODING__", :options => '-KU').should == Encoding::UTF_8.to_s
-  end
-
-  it "to Encoding::Windows_31J with -Ks" do
-    ruby_exe("print __ENCODING__", :options => '-Ks').should == Encoding::Windows_31J.to_s
-  end
-
-  it "to Encoding::Windows_31J with -KS" do
-    ruby_exe("print __ENCODING__", :options => '-KS').should == Encoding::Windows_31J.to_s
+  it "warns that -K is deprecated" do
+    ruby_exe('', :options => '-K', :args => '2>&1').should == \
+      "The -K option is deprecated. Rubinius internal encoding is always UTF-8\n"
   end
 end

spec/ruby/command_line/dash_upper_u_spec.rb

@@ -1,41 +1,6 @@
 describe "ruby -U" do
-  it "sets Encoding.default_internal to UTF-8" do
-     ruby_exe('print Encoding.default_internal.name',
-              :options => '-U').should == 'UTF-8'
-  end
-
-  it "does nothing different if specified multiple times" do
-     ruby_exe('print Encoding.default_internal.name',
-              :options => '-U -U').should == 'UTF-8'
-  end
-
-  it "is overruled by Encoding.default_internal=" do
-     ruby_exe('Encoding.default_internal="ascii"; print Encoding.default_internal.name',
-              :options => '-U').should == 'US-ASCII'
-  end
-
-  it "does not affect the default external encoding" do
-     ruby_exe('Encoding.default_external="ascii"; print Encoding.default_external.name',
-              :options => '-U').should == 'US-ASCII'
-  end
-
-  it "does not affect the source encoding" do
-     ruby_exe("print __ENCODING__.name",
-              :options => '-U -KE').should == 'EUC-JP'
-     ruby_exe("print __ENCODING__.name",
-              :options => '-KE -U').should == 'EUC-JP'
-  end
-
-  # I assume IO redirection will break on Windows...
-  it "raises a RuntimeError if used with -Eext:int" do
-    ruby_exe("p 1",
-             :options => '-U -Eascii:ascii',
-             :args => '2>&1').should =~ /RuntimeError/
-  end
-
-  it "raises a RuntimeError if used with -E:int" do
-    ruby_exe("p 1",
-             :options => '-U -E:ascii',
-             :args => '2>&1').should =~ /RuntimeError/
+  it "warns that -U is deprecated" do
+    ruby_exe('', :options => '-U', :args => '2>&1').should == \
+      "The -U option is deprecated. Rubinius internal encoding is always UTF-8\n"
   end
 end

spec/ruby/core/argf/binmode_spec.rb

@@ -45,14 +45,4 @@
       end
     end
   end
-
-  it "sets the file's encoding to ASCII-8BIT" do
-    argv [@bin_file, @file1] do
-      ARGF.binmode
-      ARGF.binmode?.should be_true
-      ARGF.gets.encoding.should == Encoding::ASCII_8BIT
-      ARGF.skip
-      ARGF.read.encoding.should == Encoding::ASCII_8BIT
-    end
-  end
 end

spec/ruby/core/argf/shared/read.rb

@@ -73,13 +73,5 @@
       Encoding.default_external = @external
       Encoding.default_internal = @internal
     end
-
-    it "reads the contents of the file with default encoding" do
-      Encoding.default_external = Encoding::ASCII_8BIT
-
-      argv [@file1_name, @file2_name] do
-        ARGF.send(@method, 4).encoding.should == Encoding::ASCII_8BIT
-      end
-    end
   end
 end

spec/ruby/core/array/shared/inspect.rb

@@ -120,12 +120,5 @@
 
       array.send(@method).encoding.name.should == "US-ASCII"
     end
-
-    it "raises if inspected result is not default external encoding" do
-      utf_16be = mock("utf_16be")
-      utf_16be.should_receive(:inspect).and_return("utf_16be".encode!(Encoding::UTF_16BE))
-
-      lambda { [utf_16be].send(@method) }.should raise_error(Encoding::CompatibilityError)
-    end
   end
 end

spec/ruby/core/dir/entries_spec.rb

@@ -48,12 +48,6 @@
     entries.first.encoding.should equal(Encoding::EUC_JP)
   end
 
-  it "returns entries transcoded to the default internal encoding" do
-    Encoding.default_internal = Encoding::EUC_KR
-    entries = Dir.entries File.join(DirSpecs.mock_dir, 'special')
-    entries.first.encoding.should equal(Encoding::EUC_KR)
-  end
-
   it "raises a SystemCallError if called with a nonexistent diretory" do
     lambda { Dir.entries DirSpecs.nonexistent }.should raise_error(SystemCallError)
   end

spec/ruby/core/encoding/compatible_spec.rb

@@ -137,18 +137,18 @@
     end
 
     it "returns the String's Encoding if it is not US-ASCII but both are ASCII only" do
-      [ [Encoding, "abc",                     Encoding::ASCII_8BIT],
+      [ [Encoding, "abc",                     Encoding::UTF_8],
         [Encoding, "abc".encode("utf-8"),     Encoding::UTF_8],
-        [Encoding, "abc".encode("euc-jp"),    Encoding::EUC_JP],
-        [Encoding, "abc".encode("shift_jis"), Encoding::Shift_JIS],
+        [Encoding, "abc".encode("euc-jp"),    Encoding::UTF_8],
+        [Encoding, "abc".encode("shift_jis"), Encoding::UTF_8],
       ].should be_computed_by(:compatible?, /abc/)
     end
 
     it "returns the String's Encoding if the String is not ASCII only" do
-      [ [Encoding, "\xff",                                  Encoding::ASCII_8BIT],
+      [ [Encoding, "\xff",                                  Encoding::UTF_8],
         [Encoding, "\u3042".encode("utf-8"),                Encoding::UTF_8],
-        [Encoding, "\xa4\xa2".force_encoding("euc-jp"),     Encoding::EUC_JP],
-        [Encoding, "\x82\xa0".force_encoding("shift_jis"),  Encoding::Shift_JIS],
+        [Encoding, "\xa4\xa2".force_encoding("euc-jp"),     Encoding::UTF_8],
+        [Encoding, "\x82\xa0".force_encoding("shift_jis"),  Encoding::UTF_8],
       ].should be_computed_by(:compatible?, /abc/)
     end
   end
@@ -161,18 +161,18 @@
       end
 
       it "returns the String's Encoding if it is not US-ASCII but both are ASCII only" do
-        [ [Encoding, "abc",                     Encoding::ASCII_8BIT],
+        [ [Encoding, "abc",                     Encoding::UTF_8],
           [Encoding, "abc".encode("utf-8"),     Encoding::UTF_8],
-          [Encoding, "abc".encode("euc-jp"),    Encoding::EUC_JP],
-          [Encoding, "abc".encode("shift_jis"), Encoding::Shift_JIS],
+          [Encoding, "abc".encode("euc-jp"),    Encoding::UTF_8],
+          [Encoding, "abc".encode("shift_jis"), Encoding::UTF_8],
         ].should be_computed_by(:compatible?, :abc)
       end
 
       it "returns the String's Encoding if the String is not ASCII only" do
-        [ [Encoding, "\xff",                                  Encoding::ASCII_8BIT],
+        [ [Encoding, "\xff",                                  Encoding::UTF_8],
           [Encoding, "\u3042".encode("utf-8"),                Encoding::UTF_8],
-          [Encoding, "\xa4\xa2".force_encoding("euc-jp"),     Encoding::EUC_JP],
-          [Encoding, "\x82\xa0".force_encoding("shift_jis"),  Encoding::Shift_JIS],
+          [Encoding, "\xa4\xa2".force_encoding("euc-jp"),     Encoding::UTF_8],
+          [Encoding, "\x82\xa0".force_encoding("shift_jis"),  Encoding::UTF_8],
         ].should be_computed_by(:compatible?, :abc)
       end
     end