# Trinity survey followup The following are some followup questions related to your initial answers to interop survey. Please add answers inline to the document. * (2.3) How's py-discv5 coming? Is it ready to run isolated tests against another language implementation? Jannik: ~~It is *almost* ready for some tests against other implementations. The only crucial thing missing is the handshake key derivation function which I've stubbed for now, but plan to implement tomorrow (should be just two lines of code of some library)~~ (done). Other than that, I took some shortcuts, so some parts are missing, but shouldn't stop us from testing it: - flat routing table, no Kademlia (doesn't really matter for small networks) - no permanent storage of ENRs or the routing table (but they can be initialized via a config file) - no majority voting on ENR endpoint, I just believe everyone who sends me a Pong - I don't request nor answer FindNode messages with non-zero distance (hope to finish this by the end of this week) Everything else seems to work, at least in local tests. * (2.7) Have you run a trinty-to-trinity testnet yet? * Sounds like no, but Alex plans to try some experiments this week before interop * (3) Any updates on syncing from genesis after the network has been running for some time? * [Alex] not sure * (5.1) Are you gossiping unaggregated attestations upon creation? Are you aggregating locally for block production? What's the status of your attestation pool? * (7.1) Do you pass the SSZ_static tests? * [Alex] we do, although it didn't surface a bug in our `py-ssz` implementation which is concerning * [hww] ^^^^^ the bug was due to cache; we found out it from Mainnet tests. * (11.1) Any update on implementing the proposed metrics? * [Alex] not sure * [HW] No * (13.2) Is the SSZ tree-hash caching sufficient to handle the minimal config 8s slot times and low validator counts? * [Alex] we seem fine running at very low validator count, hopefully can get some better numbers this week * [HWW] current numbers of benchmarking "calculating the hash tree root of the genesis_state without cache": * 1000 validators + `mainnet`: 510.64 ms (OPS: `1.95`) * 100 validators + `minimal`: 122.88 ms (OPS: `8.13`) * (15.1) Can you start from a prepared SSZ genesis state? * [Alex] i expect so, will know more before interop * Anything else you want to share or discuss before next week?