< sipa>
base_encode_decode: pubkey.cpp:234: bool CPubKey::IsFullyValid() const: Assertion `secp256k1_context_verify && "secp256k1_context_verify must be initialized to use CPubKey."' failed.
< sipa>
that looks like an initialization order issue in the fuzzer
< jnewbery>
Hi folks. I'm not going to be around at the normal time for the p2p meeting tonight (21:00 UTC). There currently aren't any suggested topics in https://github.com/bitcoin-core/bitcoin-devwiki/wiki/P2P-IRC-meetings so I suggest we just cancel this time around, but if anyone else wants to host, please go ahead!
< sipa>
i may not be around either
< gleb>
I rebased #21515 to the reviewable state again, I was going to mention that and also ask sipa about moving forward with minisketch
< gleb>
I mean, I not only rebased it but also implemented what sipa and Greg suggested, and simplifies a bunch
< sipa>
gleb: cool, did you redo simulations?
< gleb>
Yeah, my new simulator is at github.com/naumenkogs/txrelaysim. Supposedly very reproducible and easy to use
< gleb>
basically 2 inbound + 2 outbound flood has good bandwidth/latency
< gleb>
Didn’t measure “who does most tx relay work” balance, black holes and privacy yet, but that’s secondary and easy to tune I think
< gleb>
My biggest worry so far was that “2 inbound” is easy to game with sybils, but I conclude that it doesn’t harm the perf much. Also later we might select them with a heuristic, but I think not now
< gleb>
Also dropped the part with recomputing Q because static Q=0.01 works really well and complexity not needed