Eth2.0 Implementers Call 14 Agenda

[djrtwo]

Eth2.0 Implementers Call 14 Agenda

Meeting Date/Time: Thursday 2019/3/14 at 14:00 GMT

Meeting Duration 1.5 hours

YouTube Live Stream Link

Agenda

1. Testing Updates
2. Client Updates
3. Research Updates
4. quick update from raul/protocol labs
5. lighthouse benchmarks update
6. Leap seconds and time drift
7. Network spec
8. Quic(k) update
9. shuffling swap semantics
10. serialization benchmarks
11. Spec discussion
12. Open Discussion/Closing Remarks

[jrhea]

I watched Justin's talk at EthCC and he mentioned that slots will account for leap seconds. Can we talk about the reason why this is necessary? Also, I'd like to hear what other people think about the potential problem with time drift and if we plan on using some sort of method for time sync.

[zscole]

I watched Justin's talk at EthCC and he mentioned that slots will account for leap seconds. Can we talk about the reason why this is necessary? Also, I'd like to hear what other people think about the potential problem with time drift and if we plan on using some sort of method for time sync.

Yes, I have opinions.

I would also like to discuss the need for TLS support in the wire protocol. I don't think this is necessary and would only incur additional overhead.

[Mikerah]

Can we let the Pegasys team go into details about their results on using QUIC for handel? They posted a summary in this issue and I think it can help in informing us in which transports to support/use.

[bkolad]

@Mikerah thanks for bringing it, sure we can give an update about our experience with QUIC.

[nkeywal]

Can we let the Pegasys team go into details about their results on using QUIC for handel? They posted a summary in this issue and I think it can help in informing us in which transports to support/use.

We're more than happy to do a quic(k) update during the call :-)

[protolambda]

ethereum/consensus-specs#774
(TLDR: spec defines shuffling as a way to retrieve the right validators by querying for a sequential subset of the output, i.e. a committee. This is technically un-shuffling. Shuffling here is technically only really used in implementation to get the new list permutation from the old list permutation)

Discussion about semantics of shuffling, short read, just something to agree on. Would like to discuss this in the call. Vitalik also mentioned another related issue: ethereum/consensus-specs#729
(TLDR: in phase 1 we introduce a "shard" param to get a specific committee, which is also really just a sequential subset of the shuffled list.

[paulhauner]

I have been bench-marking per-block/state processing across different hardware (my laptop and desktop) with different validator counts and scenarios (full block, not-so-full block).

I don't have a whole lot to say that isn't in the doc (apart from that I am sick of markdown tables). Happy to field questions or feedback though :)

https://github.com/sigp/serenity-benches

[leobago]

I have been bench-marking per-block/state processing across different hardware (my laptop and desktop) with different validator counts and scenarios (full block, not-so-full block).

I don't have a whole lot to say that isn't in the doc (apart from that I am sick of markdown tables). Happy to field questions or feedback though :)

https://github.com/sigp/serenity-benches

It seems the links to the source codes are not working.

[paulhauner]

It seems the links to the source codes are not working.

Thanks for the heads up!

I've pulled them out for now. You can find the source here: https://github.com/sigp/lighthouse/tree/master/eth2/state_processing/src :)

Most optimized is here: https://github.com/sigp/lighthouse/tree/sane-case/eth2/state_processing/src

[djrtwo]

Quick update from Raul from Protocol Labs:

We’re working to add deprecation notices to the areas of the spec that are outdated.
Making significant strides with https://docs.libp2p.io
Writing a non-normative walkthrough of the libp2p stack that everybody can use as a reference
Engaging in various debates on GitHub

[zscole]

We're working on a lightweight wire protocol. https://github.com/Whiteblock/hobbits/blob/master/Ethereum-Wire-Protocol-0.1.EBNF