On Wednesday, February 13, 2013 at 4:05 PM, Giovanni Bajo wrote: > You probably forgot to tell your security researcher that we *start* from > sha1 hashes. > No I told him, But Richard has said he's going to do a forced password reset a week after he sends an email to everyone informing them of that. Int hat case the risk to keeping the unsalted sha1's around for another week is pretty minimal. > > bcrypt(sha1(pt)) shouldn't be used as a "final algorithm" because sha1 is > academically broken and might be real-world broken in the next few years to > the point to actually reduce entropy a bit (but let's also remember that a > normal average password has an estimated entropy in the range 20-40 bits). In > fact, nobody here is suggesting to use bcrypt(sha1(pt)) forever, and in fact > the code would upgrade to bcrypt(pt) as soon as possible (first login). > > But there is no question that it's far better to store bcrypt(sha1(pt)) in a > database rather than sha1(pt). I would be surprised if somebody argued > otherwise.
_______________________________________________ Catalog-SIG mailing list [email protected] http://mail.python.org/mailman/listinfo/catalog-sig
