release-0.3.20 plan: c20-1's kept start on the pin, the Mac build started

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-07 12:47:20 +00:00
parent e499277bc3
commit b528038e45

View file

@ -443,3 +443,5 @@ Speculative builds on the candidate from the ship worktree, box only, sequential
**c4459193's node gates, both PASS on its own binary (the node lane; built 12:33Z, sha256 45be9b02d1b002f5, string read back):** digest gate 12:33:27 to 12:35:05Z: thirteen fields a89be8a7 on both binaries, sixteen fields db9a85f9 refused with the line and no peer, the live file's digest eada4bda unmoved. Mixed-version gate 12:35:26 to 12:45:39Z: digest b0afb2ee on all five, 215 new and 314 old blocks accepted, 0 rejected, plain header version 2, counts equal at 312, 441 and 529 through both clean joins and the restart step (the new node restarted 12:43:08Z on its own datadir and resynced, where b7cc37e7 met the lock), no panic in any log. The node side's lines are complete; outstanding for the pin: the wipe canary on c19-1, the cases rerun on c20-1, the 12 GB settled-claim line.
**The Mac's binaries start now (13:4x BST):** on the node gates' pass the Mac builds its own igneumd and igneum-miner from c4459193 under the build lock, one at a time (the Mac rule), so the DMG follows the wipe line by minutes; a canary fail voids them.
**The cases rerun's target up (the fleet, 12:45:46Z):** c20-1 on c4459193 (sha 45be9b02d1b002f5 read back on the pod) started on pool-1's 0.3.17 copy with the N13 rewrite line and "state blob of layout 2 read and converted" once, panicked 0, catching up from 129,398 blocks: the pinned binary's kept-datadir read on a second pod. Relay j84mqzlai3yax9 and poison vnm1dlx8hkocq2 against it on its synced line; CASES END about 14:20Z (15:20 BST). c19-1's wipe IBD at 15 percent of the headers at 12:43Z, on the class. The Mac's node build started under the lock (scratch r0320/mac-build.sh: igneumd and igneum-miner, then the DMG).