Eth2.0 Implementers Call 12 Agenda

[djrtwo]

Eth2.0 Implementers Call 12 Agenda

Meeting Date/Time: Thursday 2019/2/14 at 14:00 GMT

Meeting Duration 1.5 hours

YouTube Live Stream Link

Agenda

1. Testing Updates
2. Client Updates
3. Research Updates
4. discv5 update/discussion
5. Quick note on light clients
6. Minimal phase 0 wire api
7. Path to testnets
   • suggested reading: https://lighthouse.sigmaprime.io/update-07.html
8. Genesis slot number & signed vs unsigned ints
9. Spec discussion
10. Open Discussion/Closing Remarks

[Mikerah]

Can we discuss everyone's expectations for testnets in March and set better deadlines for making this happen? It is getting to a point where we need to discuss during the calls and make clear our expectations to the community.

[djrtwo]

@Mikerah Yep. Also suggested reading: https://lighthouse.sigmaprime.io/update-07.html

[cemozerr]

Can we also discuss the GENESIS_SLOT start number?

[djrtwo]

@cemozerr -- this might end up being more contentious than big vs little endian!

[akhila-raju]

Just a nit - description says Call 11 rather than 12

[djrtwo]

Just added some additional notes about sync especially with respect to weak subjectivity in pos and how to handle in the sync protocol. Please review. If we have general consensus tomorrow on this API, then we can get it into a solid spec and unblock networking

ethereum/consensus-specs#593

[karalabe]

One thing I'd like to bring up is SSZ's design requirements. I've read various data pieces here and there, but the spec doesn't really say what the goals of the data format are. Those I think are essential to clearly state to allow challenging the design / iterate on it.

E.g. I have a challenger called SOS, which has O(log N) data access complexity compared to SSZ's O(N) for dynamic types (i.e. lists). https://gist.github.com/karalabe/3a25832b1413ee98daad9f0c47be3632

• If EVM accessibility is needed, SOS is much better than SSZ. But do we want EVM to extract data?
• Encoding is currently is unfeasible on the EVM (mem + proc). Do we need the EVM to encode data?
• Neither formats are stream encoders, meaning encoding shuffles a lot of memory. Is that a problem?
• The spec also uses this format as a tree hasher. Is the goal hashing and serialization less relevant?
• The idea of the tree hasher is to allow Merkle proofs. But does the scheme support partial updates?

The answers to these questions can be the deciding factors whether to favor one format over another, and I haven't really seen either of these explored much. I get it that this seems such an irrelevant tiny little detail compared to the grand scheme of things, but I'd like to remind everyone that one of the biggest scaling limits in Ethereum 1.0 is the Merkle trie's design, which haunts every subsystem.

[5chdn]

Re: (7.)

To launch a testnet, we would need to come up with some networking layer, especially discovery seems to be largely undefined. Is there anything discussed yet? For a Parity Shasper mono-client testnet, we could just reuse our Substrate network crate in the short term. Is this something that would work out for now or shall we double down on spec'ing networking?

[nkeywal]

Shall we have a workshop 8th, March, just after ethCC Paris? We can host it if people are interested.
Please thumbs up if you can make it, I will confirm (or not :-( ) tomorrow.

[paulhauner]

We'll have a team member at EthCC but they're on vacation so I'm not sure what their plans are. I'd say it's likely they'd attend an Eth2 meet.

[protolambda]

After EthCC (https://ethcc.io/, 5-7 March) there is also an ETHGlobal hackathon, ETHParis (https://ethparis.com/, 8-10 March), which I don't want to miss. Can we plan the meetup at one of the venues? Alternatively, we could also just make this part of the ETHmagicians meetup on March 4 (paid (food), same venue as EthCC, https://pretix.eu/ethmagicians/councilofparis2019/)

[nkeywal]

Ok, 6 people it's a little bit short to do a workshop. It would work for a beer however :-). Let's try to catch up during the conference, we can organize an impromptu small session like the ones we had during the devcon if necessary.

[5chdn]

Integrating it to EthMagicians makes perfect sense. Just propose a Serenity ring.

https://ethereum-magicians.org/t/council-of-paris-call-for-rings-and-participants/2588

[protolambda]

I'm up for Ethmagicians, but other options work too. And others may not already be in Paris. The beer + impromptu plan of @nkeywal has the best chance to work out I think.