以太坊研究日报
A daily record of what changed in Ethereum protocol work, research and the ecosystem.
Page 13 of 19 · August 13, 2026 to August 19, 2026
nerolation submitted an EIP-8141 PR sketching the expiry mechanism as a new frame mode instead of a special-target VERIFY frame, and Marchhill noted that installing the expiry verifier leaves its nonce unspecified. nero also wrote EIP-8131's data floor and EIP-8279's BAL floor as independent of the gas limit. Derek Chiang pointed out that Base's multi-prover bridge to Ethereum takes a day, slower than bridging to Solana. L2BEAT's privacy pages added a "note discovery" section.
lightclient added a `SIGDATACOPY` instruction to EIP-8141, splitting the copying of `ARBITRARY` signature data out of `SIGPARAM` so that `SIGPARAM`'s stack requirement becomes static, merged the same day; the August 6 initial warm set PR and the August 14 banned-opcode relaxation were also merged. chunter-cb proposed L1 and L2 adoption profiles for EIP-8130. ladislaus warned that a plain ETH transfer is no longer always 21,000 gas after Glamsterdam. In the evening FOCIL Breakout #40 set focil-devnet-0 for August 31.
The EF blog and ethPandaOps announced Platåberget, a community-facing Glamsterdam testnet meant to run for months, with Glamsterdam activating on it on August 20; the blog warns that wallets, indexers and gas estimators relying on a hardcapped maximum gas limit will break. On wallet infrastructure, leekt submitted a wallet-to-paymaster web service standard for frame transactions and `eth_estimateFrameTransactionGas`. EF Protocol Architecture published its Hegotá view: FOCIL is the right headliner and frame transactions should join it.
On Sunday night Ethlabs published a long piece rating each Hegotá proposal: EIP-8141 frame transactions, EIP-8250 keyed nonces and EIP-8272 recent roots are all A-tier, because together they let privacy protocols send private transactions through the public mempool without centralized relayers. kassandra echoed it: FOCIL is already the headliner and frame transactions should join it. nero_eth explained that the 66 proposals will be cut down over the next few core dev calls. Vitalik described Ethereum's scaling strategy as combining UTXO-style state with dynamic state.
Saturday. The EIP-8141 PR submitted on August 5 was merged: no frame in an atomic batch may declare approval scope. EIP-8130 corrected its address-derived `actorId` encoding to the right alignment used by the canonical Keystore implementation, merged. EIP-8164 now spells out what changes once an account converts. donnoh published his Hegotá ranking.
lightclient split a frame's `gas_limit` into `execution_gas_limit` and `state_gas_limit`, merged: execution gas stays a call-frame concept while state gas is handled per frame. Another set of PRs looks at migrated accounts: EIP-8151 stops a revoked ECDSA key from authorizing through `ecRecover`, but EIP-8141's `SIGPARAM` would reopen that gap elsewhere. Jochem Brouwer's benchmarks confirmed the Glamsterdam repricing numbers are safe and final. Eli Ben-Sasson criticized the EF's decision to drop Poseidon.
At ACDE #243 in the evening, client teams reached rough consensus on EIP-8141 frame transactions; L2 concerns go to a breakout first, and the next ACDE decides whether it becomes a headliner. Glamsterdam repricing numbers were finalized, making a 200M gas limit safe after the upgrade. During the day AnkushinDaniil filed a run of EIP-8141 mempool PRs: allowing several pending transactions per sender on disjoint keyed nonce sets, and skipping simulation for a canonical passkey wallet. Justin Drake announced the EF is abandoning Poseidon for L1 in favor of SHA or BLAKE.