[Python-Dev] Logging, Unicode and sockets

[MRAB]

Vinay Sajip wrote:
> Thanks to 
> 
> http://bugs.python.org/issue7077
> 
> I've noticed that the socket-based logging handlers - SocketHandler,
> DatagramHandler and SysLogHandler - aren't Unicode-aware and can break in the
> presence of Unicode messages. I'd like to fix this by giving these handlers an
> optional (encoding=None) parameter in their __init__, and then using this to
> encode on output. If no encoding is specified, is it best to use
> locale.getpreferredencoding(), sys.getdefaultencoding(),
> sys.getfilesystemencoding(), 'utf-8' or something else? On my system:
> 
>>>> sys.getdefaultencoding()
> 'ascii'
>>>> sys.getfilesystemencoding()
> 'mbcs'
>>>> locale.getpreferredencoding()
> 'cp1252'
> 
> which suggests to me that the locale.getpreferredencoding() should be the
> default. However, as I'm not a Unicode maven, any suggestions would be welcome.
> 
Well, encodings can vary from machine to machine, and if the encoding
doesn't cover all the Unicode codepoints then you could get an encoding
exception. For these reasons I'd vote for UTF-8.