Skip to content

0.5.213: adopt edge v25.2.0 / persist v44.7.0 — attribution self-refusal (#623) and the same-key rebind door (#864) - #610

Merged
emooreatx merged 2 commits into
mainfrom
feat/adopt-edge-v25.2.0
Sep 19, 2026
Merged

emooreatx merged 2 commits into
mainfrom
feat/adopt-edge-v25.2.0

Conversation

@emooreatx

Copy link
Copy Markdown
Contributor

Adopts CIRISEdge v25.2.0 / CIRISPersist v44.7.0 (verify v15.2.0 unchanged) — the #602 update. Pins only; no server behaviour changes in this cut.

What the substrate brings

What this does NOT fix

The 0.5.212 canonical serves zero rounds with zero reply-to-self failures: a first-contact peer is dropped because edge's #402 bootstrap carve-out keys on the attribution result, which is None exactly when a fresh peer needs it (#609 / CIRISEdge#624). That is an edge fix; this pin does not include it. The traceflow ladder's served stage (0.5.212) is the in-repo probe; it passes in the harness because the harness canonical admits the agent's announce before the agent dials.

Verification

  • cargo tree -i: one copy each of persist v44.7.0, edge v25.2.0, verify v15.2.0.
  • Substrate contract gate, replication policy gate, anchor-block gate: green (no ratified hash moved in v44.7.0).
  • 127-binary sweep green; lib 531; clippy --all-targets; localization guard.
  • Ladders: see the checklist comment.

🤖 Generated with Claude Code

https://claude.ai/code/session_01SZcY9yx3JZ8YqkqtFq1TQw

…bution (#623) and the same-key rebind door (#864)

Pins: persist v44.6.0 -> v44.7.0 (x4 + lens-core), edge v25.0.0 -> v25.2.0
(x2). verify v15.2.0 unchanged. One copy each; substrate contract,
replication policy and anchor-block gates green (no hash moved).

Edge v25.2.0 = CIRISEdge#623 (inbound attribution never resolves to this
node's own key; no responder is built for ourselves — the defence-in-depth
half of #607) + persist v44.7.0 (#864: Engine::rebind_key_record /
rebind_self_federation_key; edge admits the replicated Rebound). This cut
pins them. The server half of the rebind (#606) and the first-contact
carve-out (#609 / CIRISEdge#624, why the 0.5.212 canonical still serves zero
rounds) are the next cuts.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01SZcY9yx3JZ8YqkqtFq1TQw
@emooreatx emooreatx added the ci:full Run the full CI matrix (macOS + Windows) on this PR label Sep 18, 2026
@chatgpt-codex-connector

Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

@emooreatx

Copy link
Copy Markdown
Contributor Author

Local verification on the v25.2.0 / v44.7.0 wheel

  • 127-binary sweep green, lib 531, clippy --all-targets, localization guard; substrate contract / replication policy / anchor-block gates green (no hash moved).
  • Traceflow: SUCCESS incl. the served stage (harness canonical logs not priming itself, serves the agent's rounds).
  • Chat: first local run BROKEN AT sent (the joiner's KeyPackage took ~60 s to reach the creator while this box was compiling); rerun SUCCESS, cross-node chat proven. CI's chat ladder on this PR is green on an idle runner. Reading it as a load flake, not a v25.2.0 regression; noting the propagation time in case it recurs.

@emooreatx
emooreatx merged commit 1ebb7ab into main Sep 19, 2026
19 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ci:full Run the full CI matrix (macOS + Windows) on this PR

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant