Mempool and policy: ranking notes
Category notes
No PR in this category reaches P1: the strongest items are real but bounded bugs (one-block censorship window, reorg-only false eviction, operator-triggered overflow, locally-triggered crash) rather than open DoS/pinning vectors or blockers of the cluster mempool project. The band spread P2/P3/P4 is otherwise preserved as originally assessed, with one demotion (#36466).
Review order and overlapping PRs
- #36328 and #36608 (both instagibbs) both alter package acceptance/relay results; the narrow TX_RECONSIDERABLE fix (#36328) should be reviewed first, since #36608 adds 1p1c package relay on top of the same submitpackage/package-validation surface.
- #35286 (testsubmitpackage dry-run for 1p1c) and #36608 (adds 1p1c package relay to submitpackage) overlap on the package RPC surface; decide the intended package RPC shape once rather than twice — review #35286 first as it is the narrower, self-contained addition.
- #36182 and #36712 (both ismaelsadeeq) both touch fee-rate/estimation code; they are independent, but #36712's CFeeRate zero-size guard is a prerequisite-free crash fix and can land without waiting on the estimator fallback.
- #36122 self-describes as a draft accompanying BIP460/458/459 and vendors unmerged libsecp256k1 branches; its mempool-policy content cannot be settled before those upstream pieces, so it sits at the bottom of P3 on problem-certainty grounds, not on draft status.
- #36540 (removes entry-time index and age eviction) and #36214 (cluster trimming during reorgs) both exercise mempool eviction paths; if #36540 lands first the reorg/trim test expectations in #36214 may need adjusting.
Band and position changes
About
This pass by claude-opus-5 on 2026-11-01 saw every PR in the category at once and checked the bands given to each PR alone against each other, ordered the PRs, and noted chains and overlaps. Back to the category.