Configurable Salt Length for Argon2 Key Derivation

[ankane]

Hi, thanks for this awesome library 👍

Currently, Argon2 has a fixed salt length (128 bits). It sounds like this is plenty for passwords (as discussed in #509), but longer salts are needed for key derivation.

What are your thoughts on exposing this?

[mouse07410]

Could you please explain why 128 bits of salt is not enough for key derivation? Ideally, with a reference to a respected source?

[ankane]

Hey @mouse07410, here's the case I'm familiar with. Essentially, the salt acts as a secret key, and it’s desirable for the key to be more than 128-bits. Unfortunately, I haven’t been able to find much literature on it. In hindsight, it probably doesn’t make sense to add if this is the only case. I'll follow up if I find anything else significant.