EIP standards Process - 2:30pm saturday july '18


#1

The notes were much better organized, with nested hierarchy in: https://docs.google.com/document/d/1-wK7HHeupaDrRjZmeWRkzvxgEdnCGCqBhRvGAUnUu7w/edit?usp=sharing

Open questions and comments from the crowd:

  • Need for documentation
  • who has the authority?
  • LIP process for livepeer, what are the challenges that exist with EIPs? Scaling challenges? Improving signal vs noise.
  • Splitting protocol and application layer process
  • Splitting into standardization working groups
  • Scuttlebutt p2p community has no formalized process, how do we go from standard to implementation, while staying decentralized
  • How do eips decrease duplication and effort, background, how much is explicitly documented?
  • PIP wants to learn from it, plasma has things that can be added to the EIP list… what makes a good EIP?
  • Layer2 proposals
  • QIP is excited to model after EIPs… trying to be a non-org and develop a culture of the experts, to allow more people to participate.
  • Coordination scaling, social scaling, setting processes in place that have scaling in mind. How have other projects succeeded and failed?
  • IDEAS! HA! Adapted for execution. There is work that we do that could be moved upstream to EIPs, but we aren’t confident with our works

Organized notes:

Separating the EIP process into different pieces

  • Core things go to core devs
  • Python vs IETF
  • Interface RPC, contract layer, Consensus layer
  • Consensus should require pull requests to the Yellow Paper.
  • Applications should have working groups
  • Interfaces (JSON RPC) get a lot less press and
  • LAST CALL ends the process

Modeling other processes

  • Layer 2 (and other working groups) heading upstream to get their needs out of layer 1
  • For example, having chain introspection
  • Where do we draw the line?
  • We could have a nice light weight process that include working groups
  • Application layer standards often don’t need changes up streams but they piggy back off the EIP standard
  • HTTP is only a meaningful standard if people choose to use it… but its still useful to use it
  • 721 was funny, cryptokitties have a shitty standard, and then the first project deploys it with a shitty standard and the original thinkers lose their credibility in pushing the standard discussion
  • Token standards in NFTs… there is an army of people trying to standardize tokens
  • What happens when other languages come into the mix?
  • Standards use the ABI not the language.
  • We should focus on standardizing what happens on chain for data availability, then do libraries
  • There is an ODB layer, then there is the applications
  • The databases here are contracts
  • The crypto items standard 1155
  • different tokens of different types and making them all easy to group and move

Defining what makes something EIP-able

  • This is something that was better defined in the BIP process.
  • The process for contracts is different then the process for core protocol changes
  • RPC interface would be nice to standardize
  • Debug trace transaction gives you back different
  • Ethsign, private sign, both work different ways in different clients
  • Standards are really focusing on documentation
  • Does it need to come out of working groups?

Social scalability

  • Putting out requirements is really important and it should be part of the process upfront
  • Using the fellowship group to standardize might help
  • WORKING GROUPS!!
  • Token Working group, ScalingNOW! Etc.
  • Authoritative groups that have smaller scopes
  • Creating Working groups will help the process
  • Only the people that care need to watch
  • Helps organize processes around each unique groups needs
  • Editors need to certify the working group
  • Working groups can be forked
  • Everyone can follow a unique standards process
  • So you have working group last call, then you push to the editors who do the EIP last call
  • Editors become certification agencies for working groups???
  • It would be nice if working groups that are can be pointed to as the authority as types of standards
  • These are the editors
  • Greg Colvin
  • Nick Johnson (@arachnid)
  • Casey Detrio (@cdetrio)
  • Hudson Jameson (@Souptacular)
  • Vitalik Buterin (@vbuterin)
  • Nick Savers (@nicksavers)
  • Martin Becze (@wanderer)
  • How do we form working groups?
  • What are the minimum requirements?
  • No standards, but some structure would be nice
  • Working groups earn their authority and legitimacy.
  • People can circumvent it but if the important stakeholders are listening to a specific working group it will be hard for them to get their standards
  • Working groups can interrupt the final call if someone does circumvent them… you have this filibusters
  • TCR ? Signalling?
  • Experts make people shy
  • IETF is hierarchical … but can we push things out?
  • It’s important that Everyone can follow a standards process
  • There is a website !?!!? https://eips.ethereum.org/
  • EIP-1 https://eips.ethereum.org/EIPS/eip-1
  • LET’S HAVE A BLOG OFF!
  • SHOULD THINGS BE CURATED? NICK (curation) VS BORIS (accept all correctly formatted eips)?

#2

I can’t believe I agreed to a BLOG OFF…


CoinDesk article about "stuck ether", and efforts to unstuck the process
#3

hahahaha! Good luck! and sorry for the break in chatham house rules wanted to have accountability for that one ;-D


#4

We were a little loose on that (that @mattlock guy kept taking pictures!). But I think the non-digital room was good.

Good luck with your leg! But I see it is resulting in great note uploading so…


#5

hahahah it delayed the notes uploading honestly. finally getting back into the swing of things


#6

hi griff, can u expand on what u mean in these points (they are most critical and would help going forward):

  • 721 was funny, cryptokitties have a shitty standard, and then the first project deploys it with a shitty standard and the original thinkers lose their credibility in pushing the standard discussion
  • Token standards in NFTs… there is an army of people trying to standardize tokens

-greg

P.S. ello