Does state gas repricing (EIP-8037, EIP-8038) solve the state growth problem?

Protocol design · 7 participants · 11 quotes · 2026-10-23 ~ 2026-10-25

Geth's lightclients posted a public retro arguing that EIP-8037 and EIP-8038 only make it possible to raise the gas limit and do not solve the underlying problem of ever-growing state, and that breaking deployed contracts should be a last resort. Dankrad of the EF countered that pricing state more accurately is itself an improvement. Developers calling for state expiry and a process overhaul sided with lightclients.

Background

EIP-8037 and EIP-8038 propose repricing the gas costs of state access and creation, and they are scheduled for the Glamsterdam fork. They came from the view that raising the gas limit requires state access costs to reflect the actual resource consumption of nodes. In 2020, EIP-2929 raised SLOAD and CALL costs roughly threefold over the same DoS concern, and EIP-2930 provided a workaround through optional access lists for contracts that could break as a result. lightclients argues that these EIPs, like those, only have the effect of raising the gas limit and do not bind state growth itself. State expiry and EIP-8304 (trustless log index), mentioned as alternatives, try to address growing state from a different angle.

Why it matters

The outcome of this debate will decide how far gas schedule changes that break deployed contracts should be allowed, and whether to put renewed weight behind the state expiry roadmap. Watch for the results of the mainnet impact analysis of 8037 and 8038 in Glamsterdam, and whether they get postponed.

Key points

• Does repricing "bind" state growth, or does it only make a gas limit increase possible?
• Is more accurate state pricing in itself an improvement on the growth problem?
• Should we accept the UX cost of full state expiry, or live with state becoming 10 times larger?

Against — A stopgap that does not solve state growth

ً, Danno Ferrin, donnoh.gwei 💗, Kushal Babel, perper, chain native

• EIP-8037 and EIP-8038 do not solve the underlying problem of ever-growing state; they only make it possible to raise the gas limit
• The impact analysis on mainnet contracts should have been done six months ago, and breaking deployed code should be a last resort
• Instead of full in-protocol state expiry, it is more realistic to price cold state access much higher and leave DB optimization to the clients
• If a change is one whose interactions you cannot understand yourself, that is a serious warning sign that no one has proven it safe (donnoh)

For — More accurate pricing is an improvement

Dankrad Feist

• It enables more accurate pricing of state, which is nearly impossible without changing how state gas works
• It is an improvement in the sense that state can be tuned to grow less, but I acknowledge that it does not bind growth in any way
• It is easier to live with state becoming 10 times larger than to accept the UX impact of the upgrades required for full state expiry

← Debates