cfrg/draft-irtf-cfrg-hash-to-curve

two small pieces of feedback

Closed this issue · 6 comments

kwantam commented

Both from Leo as we were discussing VRF:

  1. pseudo_random_bytes is a bad name in hash_to_field, because the cryptographic notion of pseudorandomness involves a secret. Is there a better name we can use?

  2. Is it possible to remove line breaks from the test vectors in the document, to make copy-pasting easier?

I think the answer to (2) is no because we're limited to some fixed line length. @chris-wood do you know for sure?

As to (1), I don't have any ideas but I'll think about it.

chris-wood commented

Yeah, our hands are somewhat tied by the RFC editor for (2). :-\

As for (1), what do you mean by it "involves a secret"?

kwantam commented

Almost-formally, we say that a pseudorandom generator is a deterministic function G and some secret s such that G(s) is statistically indistinguishable from random to anyone who doesn't know s.

Here, there's no secret---all inputs are public---so "pseudorandom" doesn't quite apply, so we're sort of watering down / confusing terminology by using it in this way.

chris-wood commented

Okay, I thought you meant that its output is secret, or something :) I'm not sure folks will find this particularly confusing, though hedging against any risk is probably best. What about uniform_bytes?

kwantam commented

uniform_bytes sounds great!

kwantam commented

(I suppose your life would have been easier if I'd mentioned this before you put all that work into #259!!!)

  • ❤️1
chris-wood commented

Hah, no worries :-) I can submit a PR for this, unless you'd like to take a crack at it?