ethereum/sharding

https://tinyurl.com/eth2-handbook

Hsiao-Wei Wang

@hwwhww

Awesome 👏

Mikhail Kalinin

@mkalinin

wow, that's great!

Mamy Ratsimbazafy

@mratsim

in the dev handbook https://notes.ethereum.org/s/BkSZAJNwX lighthouse is called lighthoust.

Chih Cheng Liang

@ChihChengLiang

@mratsim Fixed now

Mamy Ratsimbazafy

@mratsim

:+1:

Raul Jordan

@rauljordan

yay we have a running v0.4 beacon chain with 8 validators :D

[2019-03-01 11:24:35]  INFO core/state: Previous justified epoch PreviousJustifiedEpoch=37
[2019-03-01 11:24:35]  INFO core/state: Justified epoch JustifiedEpoch=38
[2019-03-01 11:24:35]  INFO core/state: Finalized epoch FinalizedEpoch=36
[2019-03-01 11:24:35]  INFO core/state: Validator Registry Update Epoch ValidatorRegistryUpdateEpoch=36
[2019-03-01 11:24:35]  INFO core/state: Validator registry length NumValidators=8
[2019-03-01 11:24:35]  INFO core/state: Validator balances: [7069299 7039939 7055138 7070311 7055116 7069805 7024753 6824788]
[2019-03-01 11:24:35]  INFO core/state: Validator registry update epoch ValidatorRegistryUpdateEpoch=36
[2019-03-01 11:24:35]  INFO blockchain: Slot transition successfully processed slotsSinceGenesis=319
[2019-03-01 11:24:35]  INFO blockchain: Block transition successfully processed slotsSinceGenesis=319
[2019-03-01 11:24:35]  INFO blockchain: Epoch transition successfully processed SlotsSinceGenesis=319
[2019-03-01 11:24:35]  INFO operation: Attestation removed attestationRoot=0x3b1ac13feb127f180a75646323feece1d67436b2de7a045617a193d7d39f1622
[2019-03-01 11:24:35]  INFO blockchain: Chain head block and state updated blockRoot=0x341dfefe216aaeb992f37346c9f6e47859d1b4f7e72d2db64cf5bbe55267ea7e

Jacek Sieka

@arnetheduck

nice, congrats!

Danny Ryan

@djrtwo

🎉

Diederik Loerakker

@protolambda

https://twitter.com/MPtherealMVP/status/1101558183280238592?s=19

I'll be there at Eth magicians, so I can answer some questions if necessary. Anyone else want to help?

Gregory Markou

@GregTheGreek

@protolambda will also be there, glad to help

Mamy Ratsimbazafy

@mratsim

it’s on March 4? I can help though I’m not sure how to attend Eth Magicians

Diederik Loerakker

@protolambda

There should be a site with, and another where you can register, you pay for food/drinks during the day.

Here: https://ethereum-magicians.org/t/council-of-paris-2019-announcement/2438

*with info

Danny Ryan

@djrtwo

Justin and Carl both might be there. Looks like itll be a good and knowledgable crew! Sad to miss it

Paul Hauner

@paulhauner

Is it true that now that Transfers exist, anyone who deposits < MAX_DEPOSIT_AMOUNT needs to be ready to run a validator client in case someone uses a Transfer to make them eligable for activation?

Mikhail Kalinin

@mkalinin

@rauljordan

yay we have a running v0.4 beacon chain with 8 validators :D

Looks fantastic! is it a network or a local simulation?

terence tsao

@terencechain

It's a local network :)

Mikhail Kalinin

@mkalinin

so, there is a network stack in there and blocks are propagated through some wire protocol. right?

Jacek Sieka

@arnetheduck

looks like https://notes.ethereum.org is down

Gregory Markou

@GregTheGreek

@paulhauner this brings up a thought I had.. because we support transfers, exchanges can start operating if they chose to. Which means the massive wallets they hold would get slashed aggressively

Danny Ryan

@djrtwo

some 2.0 people live in paris — https://www.youtube.com/watch?v=r8235mBjHSs&feature=youtu.be

terence tsao

@terencechain

awesome!

Danny Ryan

@djrtwo

Is it true that now that Transfers exist, anyone who deposits < MAX_DEPOSIT_AMOUNT needs to be ready to run a validator client in case someone uses a Transfer to make them eligable for activation?

Yes

Raul Jordan

@rauljordan

Hey @mkalinin yep we have blocks propagated through the network, we're using libp2p with protobufs through a feed subscribe, send mechanism

Jacek Sieka

@arnetheduck

@vbuterin you mentioned storing historical states in an exponential falloff pattern - in principle it would have nice characteristics, but one would have to work around having to regenerate the states on every epoch

vbuterin

@vbuterin

Wait why on every epoch?

You would only need to recalculate state in the event on a reorg

You would store the last and second last state that has slot % 2**k = 0 for every k

So you only need to overwrite old states with new states as you progress forward

Paul Hauner

@paulhauner

Is there something that stops the replay of a Transfer inside a block? E.g., including the same transfer 16 times in a single block?

oh just saw it: "...and that all transfers are distinct"

vbuterin

@vbuterin

Yep

Note that the security of that is conditional on BLS signature non-malleability

Which is a property that BLS does have

But if we want to make the assumption safer we could make it stricter and say all transfers must have a unique signature hash

Jacek Sieka

@arnetheduck

@vbuterin what I understood the exponential falloff to be was that you have the state for epoch - 2**k for k=0..x etc available to you, so in the "normal" reorg scenario where you just missed a block due to timing, you only ever need to replay from the previous epoch - then it gets progressively longer as the reorg depth grows, balancing storage requirements and replay time for those deep reorgs where nothing is finalizing for a week - but that precise scheme requires that epoch-1, epoch-2, epoch-4 etc be stored which obviously doesn't work.

a related question is how to make sure you have a finalized state stored? you only know after the fact that a state is going to become finalized - this limits your options to either replaying from the previous finalized state when a new state is finalized, or preemptively storing every (epoch) state and marking them as finalized, when they become so?

Jacek Sieka

@arnetheduck

if I store only the last two two states whose slot % 2**k = 0, doesn't that mean that the (average) distance between current slot and last-stored-slot keeps growing?

Nishant Das

@nisdas

@vbuterin Unless I am missing something, why store historical states with an exponential fall of pattern ? You could just store the state at the last finalized epoch and the regenerate any required state at an epoch by using the saved blocks in disk

Diederik Loerakker

@protolambda

https://www.ustream.tv/channel/bS6Cnxyk8Wp

Lighthouse talk in 15 min

Age Manning

@AgeManning

Second version happening now :). First version stopped by the fire alarm :)

Cem Ozer

@cemozerr

Honest validator client spec states that the validator client should skip over the signature verification while running the state transition function. I am wondering if we also need to skip over the verification and updating of randao for validator clients during their state transition.

Wei Tang

@sorpaas

Hmm what is SLOTS_PER_HISTORICAL_ROOT? That value looks undefined.

Alex Stokes

@ralexstokes

is anyone worried about the case where there is a tie for an Eth1DataVote?

unless i missed some discussion, it seems that it is possible (albeit unlikely)

and the spec gives no guidance on how to break ties

(lexicographic by block hash?)

Where communities thrive

People

Repo info

Activity