The Java and Python clients currently assume that all POT/PO files are in UTF-8, ignoring the header fields MIME-Version/Content-Type/Content-Transfer-Encoding. These fields should be used to select the correct character encoding when reading POT/PO files. Also, when generating POT/PO files, we should specify a reasonable value. I think it's okay to output UTF-8 in all cases, as long as we put the correct encoding into the header. We should at least check for unsupported encodings when reading, and generate sensible headers when writing.
Assigning to Scrum product owner for prioritisation.
*** This bug has been marked as a duplicate of bug 748727 ***