Skip to content

Commit df97397

Browse files
dzhelezovclaude
andauthored
docs(validation): A-eth traces cell → PASS (20/20) post-#151/#153 (#156)
The wider 20×1k seeded traces confirmation on the merged-main build (#151 ancestor-error cascade / #153 root-frame gas). All 20 windows byte-identical to ponder's RPC store across logs / receipts / traces with zero trace-parity divergences — the two classes T-eth exercised (#151/INV-20, #153/INV-21) stay resolved at 1k-block windows. Only the tolerated block size-only derivation (#76/#106) and inert event-less RPC-only blocks remain. One window independently re-run fresh (Portal + RPC re-fetch) byte-identical. Parity bound + the accounts / block-interval † deviation stated candidly. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
1 parent 68b18ff commit df97397

1 file changed

Lines changed: 1 addition & 1 deletion

File tree

VALIDATION.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -205,7 +205,7 @@ the operator (see `harness/validate/README.md`).
205205
| `F-avalanche` | euler | avalanche | factory, logs, transactions, receipts | 4×50k (seeded) | **RETIRED — vacuous by design (§5.5)** — same anchored-discovery flaw as `F-polygon`; `F-full` is the sole real factory cell | `bash harness/validate/run-cell.sh F-avalanche` |
206206
| `F-polygon` | euler | polygon | factory, logs, transactions, receipts | 4×50k (seeded) | **RETIRED — vacuous by design (§5.5).** Ran 4/4 `pass=true` but **0=0 rows every table** (empty in-window factory scan, ~75 req/window) — not counted as a factory PASS | `bash harness/validate/run-cell.sh F-polygon` |
207207
| `T-eth` | traces | eth | logs, receipts, traces | 20×300 (seeded) | **PASS (20/20) on post-#151/#153 build, 2026-07-11** — was 8 PASS / 12 FAIL pre-fix; every FAIL was one of the two trace-parity classes — root-frame `gas` ([#153](../../pull/153)/INV-21) and ancestor-error `error`/`revert_reason` cascade ([#151](../../pull/151)/INV-20) — both since made byte-identical to ponder's RPC store. Re-verified 20/20 byte-identical vs each window's preserved RPC store via a free Portal-only backfill (§5.9, §5.10) | `bash harness/validate/run-cell.sh T-eth` |
208-
| `A-eth` | traces | eth | logs, receipts, traces † | 20×1k (seeded) | PENDING | `bash harness/validate/run-cell.sh A-eth` |
208+
| `A-eth` | traces | eth | logs, receipts, traces † | 20×1k (seeded) | **PASS (20/20) on post-#151/#153 build, 2026-07-11** — the wider (20×1k) traces confirmation on the fixed build: every window byte-identical to ponder's RPC store across logs / receipts / traces — and across every shared (event-bearing) block row — with **zero** trace-parity divergences (the ancestor-error cascade ([#151](../../pull/151)/INV-20) and root-frame `gas` ([#153](../../pull/153)/INV-21) classes `T-eth` exercised are all resolved), modulo only the tolerated block `size`-only derivation (#76/#106) and inert event-less RPC-only blocks. 52,140 metered requests; one window independently re-run fresh (Portal + RPC re-fetch) byte-identical. Parity is vs ponder's own RPC store (§5.9/§5.10); accounts / block-interval sources not covered (†) | `bash harness/validate/run-cell.sh A-eth` |
209209
| `A-base` | traces | base | logs, receipts, traces † | 20×1k (seeded) | PENDING | `POOL_ADDRESS=… ROUTER_ADDRESS=… bash harness/validate/run-cell.sh A-base` |
210210
| `U-eth` | univ3 | eth | factory, logs, receipts | 10×500 (seeded) | PENDING | `bash harness/validate/run-cell.sh U-eth` |
211211
| `E-eth` | euler | eth | factory, logs, receipts | ~12 edge windows (format eras, chunk-grid ±2, deploy floor, empty, frontier) | PENDING | `bash harness/validate/run-cell.sh E-eth` |

0 commit comments

Comments
 (0)