Ethereum's core developers leaned toward shipping Glamsterdam, the network's next hard fork, with its contested gas repricing unchanged on Thursday's All Core Devs Execution (ACDE) call, but left the go or no-go decision on a mid-December testnet fork for next week.

The repricing, Ethereum Improvement Proposal (EIP) 8037, raises the gas cost of creating new state such as accounts, storage slots and contract code. The Ethereum Foundation brought in auditors ChainSecurity and Dedaub to test which deployed contracts the change could break. Ethereum Foundation Dev Tooling Lead Rahul Kothari told the call the three had reached at least 50 companies affected by the upgrade's gas repricing, and suggested mid-January as the earliest mainnet date, since the high-impact projects and most medium-impact ones said they would be ready by then. All three projects rated high impact have fixed or will fix their contracts by mid-December, he said, and the medium-impact ones expect to be ready between next week and Jan. 10.

Ethlabs co-founder Ansgar Dietrichs, an EIP-8037 co-author who moderated the call, proposed aiming for a fork of the Hoodi test network in the week of Dec. 14, realistically Dec. 16 or 17, with the decision made in one week on the All Core Devs Consensus (ACDC) call. Developers scrapped the last Hoodi date, Oct. 26, after ChainSecurity flagged gas risks. Forking Hoodi this year would commit Glamsterdam to its current spec. Dietrichs invited anyone still pushing for a change to hold "a bit of a mini sprint between now and next week" and bring a concrete proposal. "I didn't necessarily see enthusiasm," he said of the compromise, but no one spoke up to veto it.

Get Unchained's crypto news in your inbox with the free Unchained Daily newsletter.

Where the Clients Stand

Marius van der Wijden, a Geth developer, said the most-favored alternative, moving gas refunds to the end of a transaction, is costly too. Under EIP-8037 as written, a transaction can make about 160 state writes within a 16 million per-transaction gas cap, he said, and without refunds inside the transaction, callers "would need to massively increase the amount of gas" they send. Even that fix has "significant drawbacks that we don't fully understand yet," he said.

Most client teams, the groups that build the competing software that runs the network's nodes, favored scheduling. Van der Wijden backed penciling in Hoodi for Geth, Nethermind said it was fine scheduling, Reth's Dragan Rakita voiced support and ethrex's Iván said on the call he was for "scheduling now, and no spec changes." Besu, which had asked on GitHub before the call to delay the fork for a narrow fix or move EIP-8037 to the following upgrade, acquiesced. Its Justin Florentine had argued that "changing the gas accounting and breaking things like try catch and Solidity is more concerning" than breaking contracts that hard-code gas prices. "We still think that we should change the spec," he said, adding that it "doesn't seem like anybody else does." Erigon's Andrew Ashikhmin spoke against shipping as is.

The date could also slip from the consensus layer, the half of Ethereum that runs proof-of-stake. Ben Edgington, speaking for that side, flagged "a consensus breaking change on the consensus layer" and warned that "whatever we decide today, it might get vetoed" next week.

Hegota Scope Reopened

Developers also reopened the scope of Hegota, the upgrade after Glamsterdam, as concerns about post-quantum (PQ) security fueled a desire to "accelerate the PQ timelines," Dietrichs said on the call. His slide listed five options, from a normal fork of 10 to 15 of roughly 20 candidate EIPs, to a smaller one under 10, to just the core bundle or core EIP for frame transactions, the planned post-quantum platform, to dropping frames for a built-in post-quantum transaction type. The last one drew "absolutely zero support" on the call, Dietrichs said.

Client teams owe short written positions by the end of next week, for discussion in two weeks. Until then, new EIPs stay off the execution-layer development network by default, and a cross-layer test network waits on consensus developers' call on Fork-Choice Enforced Inclusion Lists (FOCIL), that layer's Hegota headliner.

Unchained's Read

Next week's decision is in practice the choice of spec: Dietrichs estimated a week or two to agree on any alternative, plus about four weeks of impact analysis, which runs past the Hoodi window. Forking Hoodi this year keeps EIP-8037 as written, while waiting for that analysis would put mainnet in late February or early March at the earliest, or April or May with changes, by his estimates.

Related Listen: Should Ethereum Tax Its Apps Like a Nation? Arca's Jeff Dorman Says It Has To