Skip to content
Blockchain Posts

protocol

Two rival EIPs filed to fix Glamsterdam's gas repricing without breaking old contracts

Nov 8 drafts: Leo Glisic's EIP-8488 splits the opcode set into V1/V2; Giulio Rebuffo's draft runs two gas counters per transaction. Both target EIP-8037/8038.

by 4 min read
protocolethereumglamsterdameip

Two days after the AllCoreDevs paused the Glamsterdam fork over the gas-repricing invariants, two rival drafts were opened on ethereum/EIPs on November 8, 2026. EIP-8488 by Leo Glisic splits the instruction set into legacy and V2 opcodes; a second draft by Giulio Rebuffo (Giulio2002) runs two gas counters per transaction, one for execution under old rules and one for billing under new rules. CoinDesk's Nov 9 write-up is the first secondary coverage of both PRs.

What each proposal does

EIP-8488 — Versioned Opcodes for Gas Repricing (Glisic, PR #12638)

  • Contracts deployed before the fork keep running on the legacy opcode set at today's gas schedule.
  • Contracts deployed after the fork use V2 opcodes that carry the EIP-8037/8038 schedule.
  • DELEGATECALL_V2 requires the target to also be V2; a V2 contract cannot delegate into a V1 target without explicit coupling.
  • GAS and GASLIMIT introspection opcodes are omitted from V2, closing the main class of contracts that read their own remaining gas and branch on it.

Net effect: apps already live stay priced where they were deployed. Only apps launched after the fork pay the new rates.

Draft without a confirmed number (Rebuffo, PR #12637)

  • One transaction, two counters, both initialized from tx.gasLimit.
  • The credits counter uses the legacy schedule and governs EVM-visible execution (what CALL, CREATE and the opcodes see as remaining allowance).
  • The billing counter uses the new schedule and sets the transaction fee.
  • If billing is exhausted, the transaction aborts in a non-catchable way, even if credits remain.
  • EIP editor abcoathup has suggested the number EIP-8487; the PR file is still eip-xxxx.md at the time of writing. Rebuffo's own PR notes that no execution-client implementation or consensus-test suite has been run against this design yet.

Net effect: everyone pays the new prices, but old apps execute under the costs they were deployed against.

What they're fixing

The two EIPs target the breakages in the pair the AllCoreDevs paused on Nov 6:

  • EIP-8037 — raises the base charge for contract creation from 32,000 gas to 183,600 gas, before code storage and execution costs.
  • EIP-8038 — state-growth / cold-access repricing. CoinDesk's write-up gives no figure for the touch-cost delta; the ChainSecurity worked example cited in the Foundation's Nov 6 post put a repriced new-slot write at 110,020 gas versus 22,100 gas under current rules.

Security researcher banteg has published an ethereum/repricing-impact repository estimating roughly 200,000 affected contracts on Ethereum mainnet. That is the baseline exposure both drafts are trying to neutralise.

Numbers

- EIP-8488 author             : Leo Glisic (@lglisic60)
- EIP-8488 PR                 : ethereum/EIPs#12638
- Rival draft author          : Giulio Rebuffo (Giulio2002)
- Rival draft PR              : ethereum/EIPs#12637
- Suggested number (rival)    : EIP-8487 (not final)
- Drafts opened               : 2026-11-08 (both)
- Target EIPs                 : EIP-8037, EIP-8038
- CREATE base cost (EIP-8037) : 32,000 → 183,600 gas
- Worked SSTORE delta
  (state-gas spent, per
   ChainSecurity)             : 22,100 → 110,020 gas (~5×)
- Est. contracts affected
  (banteg's repo)             : ~200,000 (likely incomplete)
- Next private test (pre-
  drafts, Oct 29 estimate)    : 2026-11-16 → 2026-11-20

What to watch

  1. Eth Magicians threads. Both drafts need a discussion URL in the frontmatter; abcoathup has already asked Rebuffo to open one. Watch ethresear.ch and ethereum-magicians.org for the two topics.
  2. Client implementation on EIP-8488. Versioned opcodes land in the EVM interpreter — Geth, Nethermind, Besu, Reth, Erigon all have to carry the split. The compatibility surface between DELEGATECALL and DELEGATECALL_V2 will dominate review.
  3. Dual-counter semantics. Rebuffo's design routes every call through two meters. Review will focus on the abort path when billing runs out mid-sub-call and on the receipt fields needed to make fee reconstruction deterministic.
  4. AllCoreDevs call. The Foundation's Nov 6 post made clear the fork resumes only after the invariant list is cleared. Whichever of these two drafts (or a merged shape) picks up enough client-team support on the next ACD call becomes the path forward; the private test estimated by Oct 29 for Nov 16–20 predates both.

Context

Glamsterdam's gas-model rewrite is the biggest reshape of Ethereum fee mechanics since EIP-1559. The problem that forced the pause is not that the repricing is wrong — it is that Solidity patterns from 2016–2022 hard-coded gas costs that today's chain was deployed against, and many contracts cannot be upgraded. EIP-8488 keeps those contracts on their original schedule by versioning the EVM. Rebuffo's draft keeps them on their original schedule by versioning the meter, not the instruction set. Either path, if accepted, lets the Foundation's preferred resolution — a change to EIP-8037 and EIP-8038 rather than dropping them — move forward.

The two drafts are the first time in Glamsterdam's cycle that compatibility carve-outs for already-deployed code have been proposed as first-class EIPs. Prior mitigations lived in the EIPs themselves; these ones live outside them. Whether one wins or the two converge on a hybrid, Hoodi does not re-activate until that shape exists.

Sources:

Related stories