Python 2 vs Python 3: DevString with bytes

Hi all,

I noticed an intriguing behavior of PyTango running within a Python 3 environment.

Scenario

An old, existing C++ server to control a RS232 serial line. The server has a DevSerReadString method to read chars from the serial line ; Tango output data type is DevString. The equipment connected to the serial line sends binary data.

Problem

When calling the command from a Python 2 client, everything works fine -- the returned string corresponds to the bytes sent by the equipment. When calling the command from a Python 3 client, None is returned.

Minimal example to reproduce the problem

• Server in Python 2

  from tango.server import run
  from tango.server import Device
  from tango.server import attribute, command
  
  class BytesTest(Device):
      @command(dtype_out=str)
      def bytes_str1(self):
          return "binary_data\xee\xff"
  
  if __name__ == "__main__":
      run((BytesTest,))

• Python 3 client

   Python 3.7.1 (default, Dec 14 2018, 19:28:38) 
   [GCC 7.3.0] :: Anaconda, Inc. on linux
   Type "help", "copyright", "credits" or "license" for more information.
   >>> import tango; dev=tango.DeviceProxy("id00/tango/test")
   >>> print(dev.bytes_str1())
   None
   >>> 

Additional information

Of course the problem comes from bytes/unicode handling in Python. It is also probably a bad idea to use DevString for this kind of data transfer, maybe DevVarCharArray would have been better...

According to the PyTango documentation, DevString is encoded as latin-1, but it does not seem to be the case in reality.

>>> s = "binary_data\xee\xff"
>>> print(s)
'binary_dataîÿ'
>>> print(s.encode("latin-1"))
b'binary_data\xee\xff'