Eduard S.

Here are two sources of storage overhead that I can think of. # Storage of nodes in outdated paths in Arbo For every Tree.Add / Tree.Set, some of the nodes...

After a heap profile analysis, we observed 3 main sources of memory consumption: 1. badgerdb.Txn. There's a big memory usage related to WriteTxn resources that should be released after the...

Another point: the circuit parameters may require updating in the future once the chain is already started. Changing the genesis for a running chain is not desired, so we should...

In https://github.com/appliedzkp/zkevm-circuits/issues/83 we kept track of opcode implementations in the bus-mapping, but I often missed quickly knowing the status of the opcode specs and circuit implementation. I've extended the list...

Blocked until stack trie support is completed.

If we can define an interface for the MPT stack trie we will unlock #1649 and #1369 The high level idea interface of the MPT stack trie is that we...

Currently @miha-stopar is working on the Extension Node support and also fixing MPT circuit/witness gen bugs found by testing from the zk light client. This task is open for anyone...

Sounds good to me to try this change! It can clean up the dependency management, specially for the `halo2_proofs` which has a hardcoded git url and tag in many of...

On a side note, I think it would be interesting to develop a framework to write and evaluate state machine-like circuits so that we spot these underconstraints faster. In this...

Totally agree! In the frontend-backend PR I already split the create_proof into 3 parts (which are necessary for the split): https://github.com/privacy-scaling-explorations/halo2/blob/d350e727f11daf7c19dfe5ae6bddcc7b073fdf5d/halo2_proofs/src/plonk/prover.rs#L149 I think it can be split further into smaller...