[emwang-jump]

supposedly in some extremely degenerate cases + fd is at 60% of stake then ....

[github-actions]

 ┌─ ⚡ PERF · c9f1484 vs main@cd5830d ─────────────────────────────────
 │ SUITE                               BASELINE          NEW         Δ
 │ replay tps, mainnet               13,732 tps   13,654 tps  ·  -0.57%
 │ bench tps, localnet              670,939 tps  672,017 tps  ·  +0.16%
 │ snapshot load, testnet                4.97 s       4.94 s  ·  -0.71%
 │ mem total, mainnet                150.35 GiB   150.35 GiB  ·   0.00%
 │ mem total, testnet                 86.66 GiB    86.66 GiB  ·   0.00%
 │ clean compile, firedancer             3.16 s       3.20 s  ·  +1.25%
 │ binary size, firedancer             57.97 MB     57.97 MB  ·   0.00%
 ├─────────────────────────────────────────────────────────────────────
@@ 0 REGRESSIONS · 0 WARNINGS · 0 IMPROVED · 7 NOISE @@
 └─────────────────────────────────────────────────────────────────────

history · 6 pushes

 ┌─ HISTORY · Δ vs main, per push, newest first ──────────────────────────────────
 │ HEAD         TPS    BENCH     SNAP    MEM·M    MEM·T  COMPILE   BINARY
 │ c9f1484   -0.57%   +0.16%   -0.71%    0.00%    0.00%   +1.25%    0.00%
 │ dc1f320   +0.39%   -0.11%   -0.54%    0.00%    0.00%   -0.06%    0.00%
 │ e220fb5   +0.52%   -0.21%   +0.08%    0.00%    0.00%   -0.07%    0.00%
 │ 9ee554f   +0.56%   -0.16%   -0.19%    0.00%    0.00%   +1.02%    0.00%
+│ ea11c13   -0.76%   -0.16%   +0.25%    0.00%    0.00%   -1.61%    0.00%
 │ df1c983   -0.55%   +0.35%   +0.13%    0.00%    0.00%   -0.11%    0.00%
 └────────────────────────────────────────────────────────────────────────────────

[Copilot]

🟢 Approval recommended

The localized guard enforces the parent-completion invariant while preserving existing recovery behavior, with no blocking issues identified.

No open issues 🎉

What changed in this PR

This PR hardens replay handling for dead-block edge cases by tightening parent validation before processing a slot’s first FEC set.

Changes:

• Treat a parent with block_id_seen unset as unavailable, using the existing missing-parent recovery path.

File	Description
src/​discof/​replay/​fd_replay_tile.c	Adds a block_id_seen check to the parent lookup guard.

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

Cross-tile recovery-ordering changes need targeted runtime validation and human review before approval.

1 open issue

• Add regression test for missing parent FEC early child attachment · New

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

Dead-block retention changes cross-tile recovery ordering and block lifetimes, requiring runtime validation of eviction/reconsume, root advancement, and liveness under the reported 60%-stake scenario beyond the added unit assertions.

No open issues 🎉

1 resolved since last review

• Add regression test for missing parent FEC early child attachment

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

The consensus-critical retention changes affect replay/repair ordering and bank lifetimes, but existing tests cover only repair-side behavior and integration validation of actual replay eviction and draining could not be run.

No open issues 🎉

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

The consensus-sensitive repair/replay lifecycle changes need cross-tile runtime validation of asynchronous event ordering, replay-bank pruning, and hard-fork detection beyond the repair-state and queue-ordering unit tests.

No open issues 🎉

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

Dead-block retention changes recovery ordering and finalized-block handling, but tests cover only rotor state, leaving cross-tile redelivery ordering and hard-fork detection unvalidated.

No open issues 🎉

_{Review effort: Balanced}