EOF implementers call #66, February 5, 2025

Agenda

EOF Implementers Call #66 · Issue #1260 · ethereum/pm · GitHub

Agenda

  • Testing Updates

    • General EEST updates (mario)
    • Converting legacy tests to EEST (@chfast)
  • Client and Compiler Update

  • Non-specification Issues:

  • Specification Issues

    • Osaka-1 cleanup items
    • TXCREATE EIP-7873
      • Q: Do we need to block DELEGATECALL?
        • Will impact generated address namespace
      • Q: Is address hashing OK with everyone?
      • Q: Is toehold contract OK wih everyone?
      • Q: What other toehold variants for the ERC?
    • Add Metadata Section EIP-7834
    • Add EXTCODETYPE EIP-7761
    • Add EXTCODEADDRESS opcode

Notes

  • Testing

    • Mario discssed Improve generic EOF header · Issue #181 · ipsilon/eof · GitHub
      • There are some validaiton tests that don’t work running live, due to size
      • Markers will be added to mark some parameters as non-fillable
      • Execute has a hive mode, mocks the CL
    • Pawel spoke about Converting legacy tests to EEST
      • Tried to hand convert legacy tests
        • Months to a year if done by hand
      • Wrote python automated conversion wrapper
      • Some tests are obvious, some are obfuscarted
      • Discussed eof/statetest fixture
  • Client and compier updates

  • tesnet plan

    • osaka-1 is where the bulk of the spec freezes, for tooling support.
  • Spec Issues

  • container heder codes

    • large subcontainers?
    • VLQ encoding?
  • Punch list discussion

    • max stak height
    • code section type malliability fix - require full stack range to be accessed
      • has impact on register conversion logic if not fixed
      • easy to circumvent with deep swaps/dupn pop.
    • Punch list items will just be amendments to the existing EIPs.
  • TXCREATE

    • Question to 0X - what is the ideal predeploied toehold
      • ERC factories
        • not guaranteed to be there.
        • Tooling cannot assume
        • Adds complexity
        • Single toehold may be sufficient, keep the ERCs not “official”
      • EXTDELEGTECALL creates different namespace - feature, not a bug.
        • Perhaps Mandatory delegatecall
1 Like