# EIP-7708 **Date:** 2026 / 05 / 01 **Hosts:** Carson **Time:** 13:30 **Context:** Working session to resolve open questions on EIP-7708 (Value Transfer Logs) and its interaction with EIP-2780 (intrinsic gas repricing) ahead of Glamsterdam inclusion. --- ## Key Decisions ### 1. Scope of Transfer Logs — Deferred / Document Only **Decision:** Do not expand 7708 to cover priority fees, base fee burn, or withdrawals. Add a rationale section to the EIP explaining why. - **Priority fee / Coinbase payment:** Reconstructable from transaction + receipt + block header. Adding a log per transaction would significantly bloat blocks. - **Base fee burn:** Reconstructable from receipt (gas used × base fee). - **Withdrawals:** Not attached to a transaction, so no natural emission point. Already easily derivable from block execution; only ~16 per block. Adding logs would duplicate existing data. - **Recommended alternative:** Clients can expose a new aggregating RPC for consumers who want a unified view. **Action:** Carson to PR a rationale/clarification section to EIP-7708 documenting this decision so it doesn't need to be re-litigated. --- ### 2. Burn Logs — To Be Removed via New Self-Destruct EIP **Decision:** Remove burn logs from 7708 entirely, contingent on a new EIP (championed by Pavel) that modifies SELFDESTRUCT to never burn ETH. **Historical data motivating the change:** - Pre-Cancun: ~50 total burn events since Mainnet frontier. - Post-Cancun (after EIP-6780): ~50 events broken by the change, only 2 events that would actually emit a burn log under current 7708 spec. - Burn-from-EVM is effectively a spec/implementation artifact, not a real Mainnet feature. **Proposed SELFDESTRUCT semantics change:** - In the same-transaction-creation path, instead of fully deleting the account, **zero out three fields**: nonce, code, and storage. - **Balance is preserved** on the account. - If the resulting account is empty (zero balance), the existing EIP-150 "touched empty account" rule will clean it up. - If balance is non-zero, the account effectively becomes a fresh EOA-like account. - Preserves the MEV/CREATE2-redeploy pattern that motivated keeping SELFDESTRUCT in EIP-6780. **Consequence:** No burn can be triggered from EVM execution → all burn-log machinery (including the messy "unordered burn events at finalization, sorted lexicographically by address") can be removed. **Status:** New EIP not yet ready. Pavel to refine spec, present at next ACDE, then peer review. If accepted, ships in same fork as 7708 (Glamsterdam). If rejected, 7708 keeps burn logs as currently specified. **Edge case noted (Etan):** Fee recipient being self-destructed mid-block — flagged as the riskiest area for the new self-destruct semantics; needs careful test coverage. --- ### 3. Log Cost Accounting — Provisional Approach **Problem:** EIP-7708 currently emits transfer logs with no associated gas cost. This was acceptable when intrinsic gas was high, but EIP-2780's intrinsic gas reduction makes the ~1756 gas cost of a log significant. **Cost breakdown (for reference):** 375 base + 3 × 375 (indexed topics) + 32 × 8 (data) ≈ 1756 gas. **Discussion of approaches:** | Approach | Pros | Cons | |---|---|---| | Charge upfront in intrinsic gas | Simple, no OOG mid-execution | Can't know log count ahead of time | | Charge per-log as emitted (Pavel's preference) | Clean, consistent with other gas costs, simple spec | Introduces OOG edge cases at unexpected points; "untrusted call" callers can be surprised by side-effect costs from callees | | Bundle into CALL opcode cost | Avoids new explicit log charge; aligns with opcode repricing work | Couples 7708 to opcode repricing | **Provisional consensus:** Tack the transfer-log cost onto the **CALL** opcode cost as part of the broader opcode repricing — treat the log as a side effect of the call. This avoids introducing new OOG edge cases at finalization (which is implementation-hostile). **Outermost transfer log:** Discussed whether to omit it (info already available from tx fields) but no consensus to remove; current behavior (emit on any value transfer including outermost) retained. --- ### 4. EIP-2780 Status — Uncertain - Tony is reportedly arguing **against** EIP-2780, citing data suggesting intrinsic gas may need to *increase* rather than decrease. - Stefan: data may be skewed because account creation is bundled into the intrinsic cost; EIP-2780 attempts to split this out (creation vs. non-creation), so the data needs re-analysis. - Concerns also raised about a proposed SSTORE increase to 40K. - 2780's fate impacts whether log cost accounting needs to land in Glamsterdam at all. - Consensus: needs more research; defer log-cost PR pending 2780 direction. --- ## Action Items | # | Owner | Task | |---|---|---| | 1 | Carson | PR to EIP-7708 removing open questions + adding rationale section (priority fees, withdrawals, base fee burn explicitly out of scope) | | 2 | Pavel | Draft and refine new EIP for SELFDESTRUCT semantics change (zero nonce/code/storage, preserve balance); write extensive test cases; present at next ACDE | | 3 | Dan | Hold off on 2780 log-cost PR; revisit once 2780 direction is clearer; potentially close current PR | | 4 | Dan / Carson | Coordinate with Maria on repricing numbers next week | | 5 | Group | Re-evaluate log cost approach once opcode repricing direction firms up; favor CALL-cost bundling | --- ## Outcome Summary If Pavel's SELFDESTRUCT EIP is accepted alongside 7708 in Glamsterdam: - Burn logs removed from 7708 entirely. - Transaction finalization remains free of gas charges (no late-stage OOG). - Transfer log cost folded into CALL opcode repricing. - 7708 spec simplifies significantly. If rejected: 7708 ships with current burn-log semantics, log cost accounting deferred to a later fork (awkward, since it would mean adding logs in one fork and removing them in another).