Sync halo ecc 1013#35
Conversation
…verify_circuit now
…iy with zkevm bench
tests for single proof verify in chip for lookup, mul_add pass
…o2_proofs, so they can be uniformly patched at root level
constants during keygen_vk
now keygen_vk actually works for tests/mul_add_{single, aggregate}
Made changes because vkey write doesn't currently work.
now passes zkevm_bench with MockProver this resolves an issues where the final_pair computation in verify_circuit was incorrect because halo2_ecc did not detect (0,0) elliptic curve points, which should be treated as identity in the group
circuit and 1 state circuit * switched zkevm and halo2 repos to scroll-dev to get state circuit to prove correctly
74a4747 to
45d593f
Compare
* Update Github CI with `secrets.SECRET_REPO_DEPLOY_KEY`. * remove repo token * update cargo lock * switch rust to nightly Co-authored-by: zhenfei <zhenfei.zhang@hotmail.com>
* use embedded template for solidity * fix tera escape * check pairing in soli gen * add pairing check * more log * fix halo2 * log final pair check in instance * fix all... * ..
|
the diffs here is a bit strange... for example most changes in halo2-snark-aggregator-solidity/src/lib.rs seems already in main. the strange history is difficult for reviewing. We can continue development using this branch but before final review & merge we should refactor the git history |
|
other case is in halo2-snark-aggregator-solidity/src/lib.rs, 63 becomes 33 again in poseidon param.. Also in halo2-snark-aggregator-circuit/src/verify_circuit.rs |
|
btw later the deleted function |
|
seems fixed. TODOs before merge of this PR
|
fixed in c5ce8eb |
|
LGTM. I will use this branch for some real tests using common-rs next week. @JohnWick2ETH do you have time to have a look at this since it is a huge PR? |
|
sure. |
6b728b8 to
804fc20
Compare
|
This PR is not to be merged. |
No description provided.