search

GaloisInc / cryptol-specs

A central repository for specifications of cryptographic algorithms in Cryptol
BSD 3-Clause "New" or "Revised" License
30 stars 6 forks source link

Consider removing McEliece spec #73

Closed marsella closed 3 months ago

marsella commented 4 months ago

The McEliece has several (solvable!) issues; we should decide whether we want to address those issues to get the implementation up to par or remove / archive it from the specs repository.

Here's the scenario as far as I can tell (I'm new to the repo, so I might be missing some relevant context):

  1. The implementation doesn't typecheck. See #13 #14 #15
  2. Nobody has worked on the implementation in the last 5 years (modulo some typos), and the people who worked on it are no longer at Galois
  3. The implementation not checked in CI (because of (1), as far as I can tell)
  4. The implementation is in the wrong spot in the repo

These things will take a (likely) non-trivial amount of work to address, since someone will have to get up to speed on the protocol. However, I'm not convinced that this would be a worthwhile amount of work, since McEliece-KEM was not selected in NIST's post-quantum standardization effort (it is listed as a round 4 submission but not as a selected algorithm -- Crystals-Kyber was chosen instead).

I am not familiar enough with our clients' needs to make this call on my own, so this issue needs input from other people. Specifically:

  1. Is anyone currently using this or hoping to use it?
  2. If so, can we convince them to use a standardized algorithm instead?
  3. If not, is it better to delete it entirely or start an Archive directory at the top level of this repo for this and other deprecated and unsupported algorithms?
marsella commented 3 months ago

Based on offline discussion with @RyanGlScott and @mccleeary-galois, we're going to remove this.