Oh boy, I discovered two quite problematic bugs after releasing #swad 0.3:
* The #PAM checker could cause swad to deadlock under unlikely, but possible circumstances: Creating another PAM checker instance when the PAM helper process already died (or couldn't be started at all)
* The file checker had a bug of the stupid kind, it failed to authenticate users that don't have a "real name" set in the password file because it didn't correctly strip the newline following the hash in this case. 馃檲
Fixed them both now!
I think I'll do something you normally should never do: Re-roll the existing #release. It's IMHO kind of acceptable because this is still an incomplete 0.x version AND there are (as far as I know) no packagers yet.