Michal Čihař a écrit :
Hi
Dne Tue, 24 Jun 2014 16:15:55 -0400 Marc Delisle marc@infomarc.info napsal(a):
it seems that [0] which was done on QA_4_2 and released in 4.2.4, is causing problems to some users [1].
I got a bit lost in the discussion there, the bug only happens without mcrypt installed?
Or with a mismatched mcrypt library. I did not get a 100% confirmation about this, but this commit is the most likely candidate for the described behavior.
Maybe this kind of change should be postponed to 4.3 and reverted for 4.2.5?
Yes that could be an option, but still we need to fix that :-).
Fixing would be the best option, otherwise we'll have less people running 4.3.