I'm using a tracker with a Class named "it_issue". When
generating a new it_issue, it gets mailed out by
nosyreaction. If the subject contains umlauts, the
UTF-8 encoding of the subject will mangle the class name:

encoded subject:
Subject:
=?utf-8?q?[it_issue10]_Test-issue_with_umlauts:_=C3=A4=C3=B6=C3=BC=C3=9
F=C3=84=C3=96=C3=9C?=

as you can see, the utf-8 encoding replaces all spaces
with underscores. The resulting classname when decoding
the Subject results in [it issue10] with a space
instead of an underscore.
I've not found yet where roundup encodes the subject
(is this a bug in some python lib function?) any help
is appreciated, I'd submit a patch if you can give me
the general direction where to look ;-)

BTW: When somebody replies to such a mangled issue, an
exception is raised by the mail gw...

Happens with 0.8.4 here.

Logged In: YES 
user_id=34818

Found it. See attached patch.

Logged In: YES 
user_id=34818

Of course "=" also needs to be escaped. See attached patch
(relative to first one).

Logged In: YES 
user_id=34818

Added patch with doctests for decode_header and
encode_header. Apply if you like it.

Logged In: YES 
user_id=6405

Applied, with doctest converted to roundup-standard unit 
test.