On Thu, Feb 14, 2008 at 01:55:07AM +0100, Markus Liljergren wrote: > Hi Simon, > > I've been testing your patch, everything seems to work fine except that > Crypt_MD5 cannot be used. With either Crypt_MD5 or Crypt_DES set it behaves > as if Crypt_DES was set. Somewhere an incorrect value which then is passed > to the crypt-function is set, or is the library function crypt broken? Any > thoughts on this? I will probably try to fix this tomorrow, now it's time > to sleep. >
As the actual algorithm is determined by the given setting string to the crypt(3) function it might be possible you're using pure MD5 (32 hex chars) passwords instead. No pun intended but have you ruled that out already? For this case i found a patch that adds pure MD5 support... i'll merge and send a new diff if you need it. Kind regards, Simon
