in connection_fake_init()) too)

where merge controls was not called)

check passes

on Solaris, [v][sf][n]printf doesn't like NULL; on Solaris 10 strerror_r apparently returns NULL in some cases (ITS#4285)

cleanup and clarify the threading issues with back-ldap and back-meta when looping back to the same DSA (ITS#4141)

sounds like wiping the issues under the carpet, but test036 behaves much better if the target systems are less loaded

Likely should add UUID-named-entry test case to test026...

Need to rework test to check expected results on a per request
basis.