Lucas Xia
Lucas Xia
Figured it out: it comes from the bigfield constructor. That constructor add constraints when it breaks down the Fq values to bigfield limbs.
Well, the solution that avoided the bigfield constructor caused some sumcheck failure.
Duplicate of https://github.com/AztecProtocol/barretenberg/issues/1059.
I don't remember exactly what this is referring to. Should've attached it to a line in the code. Maybe @zac-williamson would know.
I would think we need to do this for all three still. They all output pairing point objects that need to be accumulated. I'm not sure how difficult it would...
also for the ultra recursive verifier. I observed a ~46k gate difference before and after poseidon2 was added, when it should be way lower.
Also would allow us to remove the prove_tube function and file.
I no longer think merging recursive and non-recursive is a good idea
Not fully addressed, but added asserts in https://github.com/AztecProtocol/aztec-packages/pull/13828 to enforce ordering and existence of certain backend public inputs, which should help a lot. Seems unlikely we will ever need to...
Closed by https://github.com/AztecProtocol/aztec-packages/pull/16243.