4 ms·

Here is my problem with all of this. PBKDF really just means "take this hash function and apply it 1000 times in succession". That seems like a very butchery st

Here is my problem with all of this. PBKDF really just means "take this hash function and apply it 1000 times in succession". That seems like a very butchery style to approach cryptography for me. bcrypt seems to lack any research verification. Its a modified blowfish, but who of us can tell what modifications are safe without making the scheme ineffective and gauge its validity?

tptacek 15y ago
PBKDF2 is more complicated than just iterating a hash 1000 times, but for most applications, iterating SHA1 many thousands of times is just fine. I'm not sure what kind of research verification you're hoping to get for a password hash. These aren't the kinds of constructions where the possibility of spending 48 hours of cluster compute time to generate a collision will actually hurt you. Sometimes, "research verification" is just a fig leaf for "too many Rails developers got uppity about using bcrypt and now I need to signal that I'm not one of them". Password hashes don't get a lot of research attention because they're pretty basic, cryptographically speaking.