From ab6043a502bebbf05b09034e2c812773f8a5d7ae Mon Sep 17 00:00:00 2001 From: igneum-labs <337424239+igneum-labs@users.noreply.github.com> Date: Mon, 5 Oct 2026 22:46:03 +0000 Subject: [PATCH] Ledger close status: round 1 merged, counts, evidence pointers, round 2 started Co-Authored-By: Claude Fable 5.1 --- docs/plans/ledger-close-status.md | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/docs/plans/ledger-close-status.md b/docs/plans/ledger-close-status.md index 6f55f41f1..67bb12f07 100644 --- a/docs/plans/ledger-close-status.md +++ b/docs/plans/ledger-close-status.md @@ -15,3 +15,17 @@ Round 1 running, four agents, disjoint ids, each on its own branch and worktree: Done by the closer: `docs/plans/ledger-close-plan.md`; `docs/plans/ledger-decisions.md` with eleven decision requests (M1 and M22, F16, X5 and X14, E14, X13, L1 L2 L4 L5, L3, G14, X29, P9, P21); the ledger entries M1, M22, F16, X5, E14, X13, L1, L3, L4, L5, P9, P21 now carry "Decision owner: the project lead" with the item number (commit c0b499d). Blocked without hardware, a customer or the public testnet, left Open with the blocker named: M8 (discrete AMD), M11 (multi-card rig, ROCm), P16 (a 12 GB card end to end), X15 (public testnet), X13 (a customer), E12's devnet half (phase 4), P22 and the P3 wrapper (phase 2). + +## 22:55 UTC, 5 October 2026: round 1 merged into `fud-close` (merges 37582f5, 33b7de3, 681c047, 44b42cb; closer's commit 4fa1473) + +Counts (166 entries): Fixed or rule written 54 (was 38). Answered 27 (was 22). Conceded and stated or terminal 52 (was 23). Decided or closed by rule 12 (was 11). Open with Decision owner the project lead 11 (M1 M22 F16 E14 X13 L1 L2 L4 L5 P21 X5, each with its paragraph in `docs/plans/ledger-decisions.md`; items 1 to 12). Open, blocked on hardware or a later phase 6 (P3 M16 P16 X15 P22 X17). Conceded, scheduled 2 (L3 D5). Open, minor 1 (E17). C4 is another agent's. + +What closed in round 1, with its evidence pointer: +- Group A (`ledger-text` d49eebb): 45 public-text entries moved to "Conceded, stated" with the litepaper or homepage location in each Status line; the gaps found at 18:20 UTC were written (M13 "about a fifth", P3 "Wrapped for light clients", M29 the app paragraph rewritten to Ember 0.3.9, E16 the 20% burn sentence, L9 the devnet line beside every download control, L2 the inducement wording removed, E17 "a million 100,000-gas calls a day", E13 the six-row route table from `docs/design/payment-routes.md`). Checked by grep on the merged tree: every cited sentence is present. +- Group B (`ledger-fork` bbce1fb, fork `ledger-fixes` 9e109c65 to bd1b676a from release-0.3.6): X20, X19, P15, M25, M28, X22 "Fixed on a branch, pending merge"; G10 rule written in spec 8.3. Suites: PC 2 jobs build-20261005-200049 (kaspa-consensus 95/0) and build-20261005-200606 (consensus-core 101/0, mining 52/0, igneum-exec 13/0, igneum-miner 17/0, p2p-flows 32/1, the one failure the new test's own arithmetic, fixed in bd1b676a and run on the Mac 33/0, labelled as a Mac run). M28's tampered pack refused in the CPU emulation; the GPU half on PC 2 is owed. Bench-log "ledger close round 1: fork fixes". +- Group C (`ledger-relay` 28c028b, 90967e3): X23 to X28, the curl part of X29, G13, G14 "Fixed on a branch, pending merge"; relay tests 47 of 47 in 6 suites plus 16 of 16 packaging checks; two new CI checks (`tools/ci/commit-tz-check.sh`, `tools/ci/curl-header-check.sh`). Owed to the project lead: the relay deployment and the secret rotation order in `relay/README.md`; the `schtasks` check on PC 1; the GitHub run of the signed-inputs workflow. Bench-log "ledger close round 1: relay and CI fixes". +- Group D (`ledger-sims` 1925db0, 6ff4648, e17f51b): M14 and F14 answered with evidence (the 50x pulse against a lagging retarget inside `finality_v2.py`, both W2 forms: weight over hash share 0.22 under rule v2 and 0.96 under Kaspa's rule in the DAA form, 0.19 under both in the median-time form, the renter never crosses a third under rule v2); M21 for the largest body the rules allow (490 KB bodies: two-hop p99 812 ms, k 5, against k 18 at D = 5 s); F20's test half (a 45% silent set through four fast-time epoch boundaries: the program advanced at each boundary on all three nodes, 0 new locks in the pause, 0 conflicting certificates, the first new lock 2 s after the silent set returned); X14 for hashing, proving and aggregation (signing needs a certificate RPC, round 2); E16 code half with one live block (version-stamped: node a24ab01a, DAA 125,005, no 0.3.10 flip inside the window); P17 reported (the RPC returns three words and `finalized` resolves to the tip; the four-state word is round 2 code); X12 logged from the 3 October record. Bench-log, six headings "ledger close round 1". + +Known class finding from round 1: the shared `build-inputs.zip` on master's tooling ran another agent's inputs under a ledger-fork job id (build-20261005-192537); the per-job zip fix is on `c4-fix` (unmerged to master) and group B used it from there. + +Round 2 starting (four agents): `ledger-design` (D6 review, X17 design, D5 schedule, the blocked items' status lines), `ledger-pc2` (M16 inline-cache kernel on PC 2's 5090, E17 draw line, P17 four-state RPC; one PC 2 job), `ledger-observer` (X14 signing concentration from block payloads, the X5/X14 observer columns on a branch), `ledger-tails` (F7 other block rates, F3 hostile-aggregator simulation, F17 spec tails).