Non-Unicode strings are silently ignored in StringFields

[inklesspen]

from protorpc import messages

class LogMessage(messages.Message):
    is_error = messages.BooleanField(1)
    # message value is silently ignored if it's a bytestring
    # make sure it is a unicode string!
    message = messages.StringField(2, required=True)

>>> LogMessage(is_error=False, message='fnord')
<LogMessage
 is_error: False>
>>> LogMessage(is_error=False, message=u'fnord')
<LogMessage
 is_error: False
 message: u'fnord'>
>>> LogMessage(is_error=False, message='fnord').check_initialized()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File ".tox/py27/lib/python2.7/site-packages/protorpc/messages.py", line 802, in check_initialized
    (type(self).__name__, name))
protorpc.messages.ValidationError: Message LogMessage is missing required field message
>>> LogMessage(is_error=False, message=u'fnord').check_initialized()
>>> LogMessage(is_error=False, message='fnord').is_initialized()
False
>>> LogMessage(is_error=False, message=u'fnord').is_initialized()
True

I have no problem with StringFields being required to be unicode strings. I do have a problem with non-unicode strings being silently ignored.

This seems to be a regression since 0.11.1; I do not experience this problem on 0.11.1 with #25 applied manually; I do experience it on master with #25 applied. (Needless to say, one cannot use this library without #25 applied anymore.)