Use new method of getting chars written for UTF8

lib/net.js

@@ -432,10 +432,7 @@ Stream.prototype._writeOut = function (data, encoding) {
     if (encoding == 'utf8' || encoding == 'utf-8') {
       // default to utf8
       bytesWritten = pool.write(data, 'utf8', pool.used);
-      // XXX Hacky way to find out the number of characters written.
-      // Waiting for a more optimal way: http://codereview.chromium.org/1539013
-      var _s = pool.toString('utf8', pool.used, pool.used + bytesWritten);
-      charsWritten = _s.length;
+      charsWritten = Buffer._charsWritten;
     } else {
       bytesWritten = pool.write(data, encoding, pool.used);
       charsWritten = bytesWritten;

src/node_buffer.cc

@@ -38,6 +38,7 @@ using namespace v8;
 
 
 static Persistent<String> length_symbol;
+static Persistent<String> chars_written_sym;
 Persistent<FunctionTemplate> Buffer::constructor_template;
 
 
@@ -308,11 +309,16 @@ Handle<Value> Buffer::Utf8Write(const Arguments &args) {
 
   const char *p = buffer->data() + offset;
 
+  int char_written;
+
   int written = s->WriteUtf8((char*)p,
                              buffer->length_ - offset,
-                             NULL,
+                             &char_written,
                              String::HINT_MANY_WRITES_EXPECTED);
 
+  constructor_template->GetFunction()->Set(chars_written_sym,
+                                           Integer::New(char_written));
+
   if (written > 0 && p[written-1] == '\0') written--;
 
   return scope.Close(Integer::New(written));
@@ -463,6 +469,7 @@ void Buffer::Initialize(Handle<Object> target) {
   HandleScope scope;
 
   length_symbol = Persistent<String>::New(String::NewSymbol("length"));
+  chars_written_sym = Persistent<String>::New(String::NewSymbol("_charsWritten"));
 
   Local<FunctionTemplate> t = FunctionTemplate::New(Buffer::New);
   constructor_template = Persistent<FunctionTemplate>::New(t);