Please ask on the ML for advise.

I don't see that as a bug. The created conf file depends on your local
installation and thus may or may not include a keyserver option.

I found the work-around:

Could be a gpg-agent problem.
If I don't start a gpg-agent and run the export, it will work.
If I start a gpg-agent before, it won't work and pinentry will run amok.

This has already been fixed in the trunk. I just backported it to 1.4 and 2.0.
(svn rev 5361).

This a feature and not a bug ;-)

... and I found --allow-freeform-uid, so its not stopping me from using gpg
anymore. So! Uncritting.

We have no way to merge secret keys. Thus fixing this would be very hard.
There are a couple of other problems related to this and we won't fix them either.

You are a provisional user and thus only abale to edit your own entries. We
need to run it this way to avoid trolls closing or commenting bugs. I'll change
your role now.

It appears that gnupg does not prefer those keys based on their age, but rather
that it prefers them because they were loaded into the local keyring first. if
you remove both keys, then re-add them in the opposite order, gpg seems to
prefer a different one... that still doesn't make it any better.

Fixed in svn 5345. Slightly different patch to avoid defining a symbol from the
GNUC name space.

gpg-agent --version
gpg-agent (GnuPG) 2.0.14
libgcrypt 1.4.4
Copyright (C) 2009 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later http://gnu.org/licenses/gpl.html
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.

It doesn't work with both gnupg version if they are connecting via scdaemon/pcscd.

Version numbers of gpg, gpg-agent and scdaemon are ...

Please read the gpgv man page. Hint: Look at the first error message.

Your logs show /dev/pts/7 and as I wrote in T1203:
other bug reports indicate that any /dev/pts/(single-digit) exposes the problem.

Might be related to T1203 (gpg-agent/pinentry does work on certain ttys
(/dev/pts/?)) so it should be checked if the number of tty has an influence
on the symptoms.

It turned out to be some work to reproduce the case. The length of the output
depends on the key and other factors. I created a test case which is now used
in armor.test which clearly shows this.

Note: The bug in the card's firmware has been fixed for some time now.

ping. Is that bug still relevant?

Solved. See T1214

Duplicate of T1214

Solved in a different way. See bug#1214

Duplicate of T1214

Fixed. Thanks.

David already fixed that in 1.4.
Now fixed for 2.0 too.

Marcus: I have an account on this machine. Let me know if I shall run a test
for you.

Changed the test framework again. Now it works on the buildfarm.

Tetsing on Solaris showed the problem. Now fixed in SVN.