Description of problem: When getting messages from one of my UTF-8 python files, xgettext apparently doesn't use UTF-8, but some ISO encoding when dealing with a copyright sign: This is the msgid as put into system-config-services.pot by xgettext: msgid "" "System Services Configuration Tool VERSION\n" " Copyright © 2002-2006 Red Hat, Inc.\n" [...] This is how it should be: msgid "" "System Services Configuration Tool VERSION\n" " Copyright © 2002-2006 Red Hat, Inc.\n" [...] Version-Release number of selected component (if applicable): gettext-0.14.6-4.fc6 How reproducible: Easily Steps to Reproduce: 1. Check out system-config-services from elvis 2. cd po/ 3. make update-po or: 1. create small python file (marked with "coding: utf-8") with a translatable copyright sign 2. run "xgettext --default-domain=system-config-services --add-comments --from-code=utf-8 --keyword=_ --keyword=N_ ..." on that file Actual results: See above ex.1 Expected results: See above ex.2 Additional info: I just made a test build of 0.16.1 from Rawhide and with its xgettext the UTF-8 copyright sign was left intact.
See bug #232809 for background.
Hmm I see: guess it is necessary to track down in which version of gettext this was fixed.
I doubt there will be a chance to fix this before fc7 moves to EOL, unless someone comes with a patch. So closing for now.
(Just for the record, F7 already ships with the newer 0.16.1.)