charset weirdness with non-ascii User IDs

[Daniel Kahn Gillmor]

Hi folks--

i'm seeing some strange behavior with the keyservers on GNU/Linux
systems that don't have a UTF-8 locale, or when LANG is set to something
non-UTF8:

0 dkg at pip:~$ LANG=C gpg --keyserver keys.mayfirst.org --search '=Andrew Lee (李 健秋) <ajqlee at debian.org>'
gpg: searching for "=Andrew Lee (æå¥ç§) <ajqlee at debian.org>" from hkp server keys.mayfirst.org
(1)	Andrew Lee <andrew at linux.org.tw>
	Andrew Lee (\xe6\x9d\x8e\xe5\x81\xa5\xe7\xa7\x8b) <ajqlee at debian.org>
	Andrew Lee (\xe6\x9d\x8e\xe5\x81\xa5\xe7\xa7\x8b) <andrew at debian.org.t
	Andrew Lee (§?î) <andrew at linux.org.tw>
	  1024 bit DSA key 0xB6250985, created: 2004-11-02
Keys 1-1 of 1 for "=Andrew Lee (李健秋) <ajqlee at debian.org>".  Enter number(s), N)ext, or Q)uit > q
0 dkg at pip:~$ LANG=C gpg --keyserver keys.mayfirst.org --search '=Antoine Beaupré (work) <anarcat at koumbit.org>''
gpg: searching for "=Antoine Beaupré (work) <anarcat at koumbit.org>" from hkp server keys.mayfirst.org
gpg: key "=Antoine Beaupré (work) <anarcat at koumbit.org>" not found on keyserver
0 dkg at pip:~$ 

Note that the --search for Andrew's UTF-8 User ID succeeds, but
Antoine's fails.

This behavior happens on both gpg 1.4.12 and 2.0.19, and it happens with
or without debian's gnupg-curl packages installed.

Given that User IDs must be UTF-8-encoded, i'm not sure what the right
thing to do is here.

I tried searching for this bug on https://bugs.g10code.com, but i'm
getting an error when i search for the term "charset" for some reason.

Any suggestions for the right direction for a fix?

        --dkg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 965 bytes
Desc: not available
URL: </pipermail/attachments/20120724/d099cbac/attachment-0001.pgp>