Bug#717314: gnutls priority string regression in 3.2.2

[Stefan Bühler]

Package: libgnutls28
Version: 3.2.2-1

Hi,

gnutls 3.2.2 tried to fix a bug; when removing an algorithm the list
order wasn't preserved.
but the fix introduced another bug:
"NONE:+MAC-ALL:-SHA1:+SHA1" misses SHA1 and has MD5 twice

See upstream commit:
http://gitorious.org/gnutls/gnutls/commit/913f03ccfafc37277f0a88287d02cdbb9bbfb652

So it'd be nice if 3.2.2-1 didn't migrate to testing :)

regards,
Stefan