ethereum/AllCoreDevs

Eth-Gitter-Bridge

@Eth-Gitter-Bridge

<M H Swende (holiman)> When you said it needs to be benchmarked (1380), you said no special path for precompiles

Alex Beregszaszi

@axic

I’m getting confused by all these EIPs tbh.

Eth-Gitter-Bridge

@Eth-Gitter-Bridge

<M H Swende (holiman)> Yes, you're not alone :)

Martin Holst Swende

@holiman

Hey! Since I love to audit consensus code side by side, I made this tracker-tracker, trying to locate each implementation of the eips: https://notes.ethereum.org/@holiman/SyT_rGjNr . Anyone is free to take it and turn into a ticket or issue or wiki or whatever, I'm sure there are better forms for it which makes it easier for other people to help keep it up to date. In the meanwhile, please ping me for updates

Martin Holst Swende

@holiman

One note, which I already made for aleth, but I see now is applicable for other nodes aswell: Regarding implementing calldata reduction, there's a minor snag with transaction pools -- if they start accepting istanbul-priced txs too early, that's a DoS vector. So in geth we made the pools stateful, and when istanbul hits we flip the switch to use another formula to calculate the intrinsic gas (and thus validity of) a transaction.

Martin Holst Swende

@holiman

I did some benchmarking related to EIP-1380 (lower call to self from 700 to 40). Results: https://gist.github.com/holiman/0662916aab57fb9a3b5d74703c0620cd . Follow-up over at FEM

Andrew Redden

@androolloyd

Fascinating!

Neville Grech

@nevillegrech

Hi all, I'm referring to the static gas cost analysis of EIP1884 I wrote and applied to the currently deployed contracts on mainnet (affected contracts https://contract-library.com/?s=ether&w=FALLBACK_WILL_FAIL). If the community can tolerate a few fallback functions failing, and is open to adjust the gas cost of other instructions to compensate (such as LOGX, etc.), we have the right tool to search many different combinations of these budgeting options and find a configuration that would minimize the number of fallback functions not being able to be executed with the 2300 gas budget.

Hudson Jameson

@Souptacular

@/all We have an Ethereum Core Devs meeting tomorrow (Friday) at 14:00 UTC. ethereum/pm#121

If no one from your team can make it please update us in a comment on the ethereum/pm#121 your progress implementing EIPs for Constantinople.

Hudson Jameson

@Souptacular

@/all Core dev meeting in 9.5 hours.

Agenda: ethereum/pm#121

Don't party too hard in Berlin! :tada:

Dmitriy Ryajov

@dryajov

This message was deleted

Hudson Jameson

@Souptacular

Hey Istanbul II has been renamed Berlin amirite?

Checked the notes. We are calling the fork after Istanbul for "Berlin"

Martin Holst Swende

@holiman

yes

Brett Robertson

@DBrettRobertson

@Souptacular. Concur with @holiman. From the last Core Dev meeting 68 it was decided that the Hardfork formally known as Instabul II is now called Berlin.

Tomasz Kajetan Stańczak

@tkstanczak

@holiman @karalabe is this the finally accepted wording of 2200?

https://gist.github.com/karalabe/adc43c07db9f03be82093cd5466562b0

Eth-Gitter-Bridge

@Eth-Gitter-Bridge

<Péter Szilágyi (karalabe)> No, we decided against Alexey's version last week

<Péter Szilágyi (karalabe)> So we go with Wei's original EIP

<Péter Szilágyi (karalabe)> It just needs some love text wise

Tomasz Kajetan Stańczak

@tkstanczak

this one has Wei as one of the authors, is there a link to Wei's EIP spec?

this?

@chfast The original text of the EIP is If gasleft is less than or equal to 2300, fail the current call frame with 'out of gas' exception.

Hudson Jameson

@Souptacular

@/all Meeting in 10 minutes!

Agenda: ethereum/pm#121

Join URL: https://zoom.us/j/283714012

Chat Box: https://www.youtube.com/live_chat?v=yO0WdT-J64w&is_popout=1

Alex Beregszaszi

@axic

Made some calculations of Equihash using the blake2 precompile: https://github.com/ethereum/EIPs/issues/152#issuecomment-524324077

Added this as a comment to the meeting (https://github.com/ethereum/pm/issues/121#issuecomment-524325234), if anyone wants to discuss it. However I cannot join the call for the most part, potentially around the last 30 mins of it.

Alex Beregszaszi

@axic

Another interesting bit: current zcash block height is 590338. Lets take a very optimistic estimate and say verifying one block takes 500.000 gas. In this case it would take 295.169.000.000 gas to validate the zcash blockchain on Ethereum.

Q1: would anyone pay that cost?
Q2: can zcash be validated on ethereum without validating from the genesis?

Hudson Jameson

@Souptacular

@axic Should I tag some Starkware people in here for comment on this? Specifically Q2

Tim Beiko

@timbeiko

That’s 2-3000ETH at current gas prices, @axic. Half a million dollars to verify it :moneybag:

Alex Beregszaszi

@axic

With 1 gwei gas cost that is ~296 Eth worth of gas. Which is not terribly much given what it is for, but wonder how many blocks that would take.

@Souptacular sure! We could also try to document most of this on the EIP discussion URL (ethereum/EIPs#152)

Hudson Jameson

@Souptacular

Yeah discussion URL would be great. I'll tag someone too. @EliBenSasson_twitter Could you tag others from your team who may be able to comment? We also welcome your thoughts.

GuthL

@GuthL

@axic Should I tag some Starkware people in here for comment on this? Specifically Q2

Sorry, I could not be on the call today but let me run it internally.

Do we have the price for the precompile yet?

GuthL

@GuthL

Also, is the precompile ready to provide digest of various size? (Blake2 is supposed to do this if I’m not mistaken)

Danno Ferrin

@shemnon

I think the commited EIP is 1 gas per round, in addition to the precompiled contract call cost.

Eth-Gitter-Bridge

@Eth-Gitter-Bridge

<Péter Szilágyi (karalabe)> The precompile is only the F compression function, not the full hash

<Péter Szilágyi (karalabe)> You can build all the varioud hash sizes on top

GuthL

@GuthL

Thanks for the info

Danno Ferrin

@shemnon

Follow up on the TransactionTests from the ACD. The PR only has the “Fillers” or source and are not the filled or compiled versions of the tests. Pantheon runs off the the filled versions. Retesteth currently does not do transaciton tests. I’ll see if I can get testeth up and running and get a filled version of those in a PR.

Eth-Gitter-Bridge

@Eth-Gitter-Bridge

<M H Swende (holiman)> The filled ones are in my Istanbultests repo

<M H Swende (holiman)> @shemnon

Danno Ferrin

@shemnon

PAntheon passes (after some formatting tweaks in the JSON)

jochem-brouwer

@jochem-brouwer

I don't know if this is the right channel for it, but I'm mainly interested in the design for EVM when an INVALID instruction is reached. Why would this consume all gas? Isn't the idea of gas that we "pay" for execution? Looking up if an opcode is a valid opcode and if not terminate/revert the transaction should cost (for execution) something like 3 gas..? Why is this not the case, and what is the reasonable behind it?

Nick Savers

@nicksavers

@jochem-brouwer https://eips.ethereum.org/EIPS/eip-141

Nick Savers

@nicksavers

ethereum/EIPs#141

jochem-brouwer

@jochem-brouwer

@nicksavers Thank you but this does not answer my question. This opcode is simply designated as INVALID per this EIP, e.g. it will never have any non-INVALID code per this EIP. My question is why INVALID burns all gas and not a small amount like the REVERT opcode

Bryant Eisenbach

@fubuloubu

Because that was the original behavior

Alex Beregszaszi

@axic

Changing it now depends on first introducing transaction result states properly in the transaction receipt. Currently “invalid” and “revert” are distinguished by gas usage by clients..

jochem-brouwer

@jochem-brouwer

Yes I understand this is the original behavior. I am just wondering why this was designed like this. It does not seem consistent.

Revert is a designated operation. It is not invalid. It also puts data in return data.

Where communities thrive

People

Repo info

Activity