Antithesis reports reproducing the SQLite WAL-Reset bug in 15 minutes and says 3.51.3 closes it

Sources: primary, discussion
Summary: Antithesis reproduced the SQLite WAL-Reset bug against 3.51.2 using a generic concurrent write and checkpoint workload with generic assertions, and reports the same workload comes back clean on 3.51.3. The 15-minute detection time and the tooling claims are the vendor's own, and the post closes on a sales pitch for that vendor.
Why it matters: This archive carried the Tailscale account on 2026-08-12, where isolating the bug took months of live forensics, and this is an independent reproduction of the same defect.