Why can I not use umlauts in my commit messages?

Are you using the cygwin version of bzr?

It works pretty much everywhere else. The problem is that Cygwin itself is not able to understand the terminal encoding.

Alternatively, you may be on another platform and just not have your encoding defined. For example, you may try:

export LANG=en_US.UTF-8

and then see if that allows you to use extended characters.

Basically, there are lots of ways that systems can define their encoding (iso-8859-1, utf-8, cp1251, etc), so bzr uses the environment to determine how to decode your message. If the environment isn't configured correctly, or if you are using cygwin (which doesn't support anything but ASCII) then we have this problem.

If you *are* using cygwin, then I would recommend just switching to the win32-native installers available at:
https://launchpad.net/bzr/1.7/1.7.1 (or 1.8/1.8, or the new 1.9/1.9rc1 if you want to try it).

When this error occured I was using Ubuntu Linux.

echo $LANG
de_DE.UTF-8

This seems to work for me

mbp@lithe% LANG=de_DE.UTF-8 bzr ci -m "ä" --unchanged
Committing to: /tmp/a/
Committed revision 2.

However, I see this message is raised from the code that reads in a commit message from $EDITOR, when it can't be read in osutils.get_user_encoding. I can also commit that text from vim run as the editor.

So in the first place I would suspect that Tim's editor is not actually saving the file as utf-8. Tim, what editor are you using? Can you save a copy of the message template and attach that to this bug?

This question was expired because it remained in the 'Needs information' state without activity for the last 15 days.