Batch ZK proofs for PPOPRF #207
Conversation
There was a problem hiding this comment.
Not commenting on the algorithm, here, just some suggestions for code cleanup. I ran cargo fmt to address ci lints from there. Please address the cargo clippy lints. I've added comments near a few which hopefully explain a bit what's going on with the lint. If you don't want the practice, let me know and I can address them quickly. You can also use cargo clippy --fix or the fix action in your editor if you have that set up.
The rust String type is explicitly for UTF-8 text and can't be used for constructing binary data. Look for equivalents of python list methods rather than format string equivalents, convenient as they are!
claucece
left a comment
claucece
left a comment
There was a problem hiding this comment.
It looks good! Need some changes and testing ;)
rillian
left a comment
rillian
left a comment
There was a problem hiding this comment.
Thanks for addressing earlier comments, @ankeleralph . I think it's ready to go as far as the implementation with the remaining nits addressed. That just leaves the question from @claucece about whether the ristretto mod is necessary.
claucece
left a comment
claucece
left a comment
There was a problem hiding this comment.
I think we can simplify this PR by quite a lot. There is a lot of code repetition here. We can basically use the same code of the proof generation without batching (https://github.com/brave/sta-rs/blob/main/ppoprf/src/ppoprf.rs#L48) and the verification of the proof without batching (https://github.com/brave/sta-rs/blob/main/ppoprf/src/ppoprf.rs#L71). Perhaps, even just extending those functions to take a param to batch when it is needed. Else, the two places are basically doing the same but calling different functions. Happy to have a call to explain.
ankeleralph
changed the title
This PR fixes issue #82 by batching multiple client queries for the DLEQ ZKPs following the pseudo code from draft-irtf-cfrg-voprf.