Things that need to be done
- decode user provided log messages from locale encoded byte strings to unicode strings
- decode logfile if --logfile option used
- decode the message specified in command line if --message option used
- decode edited file otherwise
- use terminal encoding to display those messages (provided as unicode strings)
- encode log messages (provided as unicode string) in UTF-8 when storing
- decode log messages from UTF-8 to unicode string when retrieving
Legacy repositories
Something has to be done with repositories having changelog messages encoded in Latin-1 or other encodings. [http://www.kernel.org/hg/linux-2.6/ Linux kenel tree] is an example. The options are
- allow users to specify repository changelog encoding in hgrc
- provide a tool to convert repositories from legacy encoding to UTF-8