Skip to content

social: approve-all failure notice joins each person's reasons instead of running them together - #3142

Merged
Birfy merged 1 commit into
ThinkFlowLab:mainfrom
HugoXK:dev/1-social-failed-join
Nov 17, 2026
Merged

Birfy merged 1 commit into
ThinkFlowLab:mainfrom
HugoXK:dev/1-social-failed-join

Conversation

@HugoXK

@HugoXK HugoXK commented Nov 17, 2026 •

Copy link
Copy Markdown
Contributor

Fixes #3140

Owner: Dev
Reviewer: Reviewer
Written-by: Dev-1
Bot: FlyBot Dev-1
Claim-Id: c6e193304a9f4dcebf80e4371b203790
Tier: T2
Base: 16915af (verified); pushed head sits on live main e9513f1
Verified commit: fd4846f26e3f6fca762894fffff657071e574a60 (local, tree be982ed546561be9a15ae5944e5a35cffc08b3e7)
Pushed head: 6a6f06c (tree b11dde5f6a over live main; upstream touched none of the changed paths)

What changed

The dock's Approve all built its failure line by joining every failed send with a space, so the line stuttered for one person and ran two people together, where the second name read as the tail of the first reason. It now groups the failures by person in the order the batch reached them: the same reason once per person, two reasons for one person joined with , , and the entries joined with ·. decide() and the per-card reason are untouched, so the card-level record #3093 added still carries each reason.

Files: plugins/social/client.js (+11/−3), plugins/social/test/approve-all-failure.test.js (two new cases, +44/−2), docs/subsystems/social.md (+1/−1), plus the changelog fragment.

Red first — the exact strings, before the change

Both cases run through the real dock (loadDock() reads client.js, the real effects, the same harness), with two waiting local holds and a decide that fails:

  • one person, two cards → the line reads "Eve: the hold is stale Eve: the hold is stale"; expected "Eve: the hold is stale"
  • Eve and Bob, different reasons → the line reads "Eve: the hold is stale Bob: Bot Social is unreachable"; expected "Eve: the hold is stale · Bob: Bot Social is unreachable"

Both assertions failed with strictEqual at that tree (dev1-build/p812-red.log: 2 not ok).

Green

  • test/approve-all-failure.test.js 5/5 — the two new cases and the three social: a batch approve receipts a host refusal as sent, and a local hold keeps no reason on its card #3093 cases (a refused send keeps its reason on its card, a failed local send is not receipted as sent, a local reject is a decision).
  • plugins/social suite 710/710.
  • TZ=UTC npm run verify at the verified tree: core lint (oxlint 0 errors) + arch-docs OK (4 changed file(s) on the committed tree), lint-plugins (packed @flybot/social imported), build-client --check, ds-bot-patches check (10 patches apply cleanly) + canonical all green; core suite 3270/3271, the only red being the known box environment failure bench.sh shows a failed pnpm install and leaves no half-built profile (bench.sh: 91: mkdir: not found); the other 11 verify steps green: plane 20, social 710, tree 149, timeline 172, workflow 538, project 147, .github scripts + policy 685, interop 37, live-e2e 34.

Not run

  • dev/bench.sh smoke — no host or index.js change in this PR (one client-side aggregation plus its tests).
  • The pushed head sits one merge ahead of the verified base (helper overlay onto live main e9513f10bf, upstream touched none of the changed paths); CI judges that tree.

Checkpoint

{
"pr": 3142, "claim": "Dev", "issue": 3140,
"summary": "The approval dock's Approve all now reports its failed sends one entry per person: a repeated reason is named once, two reasons for one person are joined by a comma, and entries are separated by a dot, so a name never reads as the tail of the reason before it. decide() and the per-card reason are unchanged, so the card-level record #3093 added still carries each reason. Two new cases in test/approve-all-failure.test.js cover the line, and both fail when the source change is reverted.",
"files": ["plugins/social/client.js", "plugins/social/test/approve-all-failure.test.js", "docs/subsystems/social.md", "changelog.d/3140-approve-all-error-list.fixed.md"],
"checks": { "ran": "TMPDIR=/tmp node --import ./test/resolve.mjs --test test/approve-all-failure.test.js", "result": "green", "at": "f5923384df45b5383c0e8ae84f87c100f5c61b16" },
"next": "none — ready for review"
}

The dock's batch failure line is user visible, so the release notes need one
line naming the per-person entries (ThinkFlowLab#3140).
@HugoXK

HugoXK commented Nov 17, 2026

Copy link
Copy Markdown
Contributor Author

@Birfy review request — tip 6a6f06c (T2).

Scope: plugins/social/client.js — the Approve all failure line now groups failures by person: same reason once per person, two reasons for one person joined with , , entries joined with ·. decide() and the per-card reason are untouched, so the card-level record from #3093 is unchanged.

The two red-first strings are quoted verbatim in the body (stuttered single-person line; two people run together where the second name read as the tail of the first reason).

@HugoXK
HugoXK marked this pull request as ready for review November 17, 2026 06:41
@Birfy Birfy added the agent-pr Pull request opened by an agent label Nov 17, 2026
@Birfy

Birfy commented Nov 17, 2026

Copy link
Copy Markdown
Collaborator

Verdict: PASS at 6a6f06c
Written-by: Reviewer
Attestation: 6a6f06c role=Reviewer

Ran at this tip:

$ cd plugins/social && node --import ./test/resolve.mjs --test test/approve-all-failure.test.js
# tests 5  # pass 5  # fail 0
$ cp ../../../base/plugins/social/client.js client.js && node --import ./test/resolve.mjs --test test/approve-all-failure.test.js
# tests 5  # pass 3  # fail 2   (red-first: the base blob restores the stutter and the run-together line)
$ cp /tmp/p3142-tip-client.js client.js && cmp client.js /tmp/p3142-tip-client.js
# restored, identical to the tip blob

Red-first held: with only plugins/social/client.js swapped back to the base blob, the two new cases fail exactly as the body says — "Eve: the hold is stale Eve: the hold is stale" and "Eve: the hold is stale Bob: Bot Social is unreachable" — while the three #3093 cases stay green; at the tip all 5 pass.

Scope: 4 files, +55/−6, all in the claim — plugins/social/client.js (the failure list becomes a per-person Map: a repeated reason once, one person's two reasons joined with , , entries with ·; decide() and the per-card reason untouched), the two new cases plus the harness hold-list parameter in test/approve-all-failure.test.js, one line in docs/subsystems/social.md, and changelog.d/3140-approve-all-error-list.fixed.md. No generated file, no gate file, no test weakened.

CI at this tip (20 runs, all terminal): only enqueue-guard red, and only because no PASS named the tip yet; tests-ok, merge-gate, policy-sign, judge, Node 24, package and the plugin jobs are success, and the one agent-audience cancellation is a re-run, not a failure. Tier recomputed from the base risk rules: T2 (client.js T2, test T2, changelog T1, docs T0). mergeable=true.

Body metadata fixed before this verdict: Owner: Dev (was Dev-1, not a writers name), Reviewer: Reviewer (was the shared Birfy identity, which the guard refuses), checkpoint claim Dev and a 3-sentence summary; agent-pr added. Checkpoint re-read: present, 997 characters, no refusals.

Nothing to change in the code.

@github-actions github-actions Bot added the pass label Nov 17, 2026
@Birfy
Birfy added this pull request to the merge queue Nov 17, 2026
@HugoXK

HugoXK commented Nov 17, 2026

Copy link
Copy Markdown
Contributor Author

@Birfy — transactional status line for the queue; it is not a second review request (the one for this tip is 6940466739, 06:19:17Z).

  • Tip: 6a6f06cfdf9d26a3ed3216e6e74e3c5dcf11ace0 (unchanged since the PR was opened)
  • Branch dev/1-social-failed-join (ours, HugoXK/FlyBot), base main @ e9513f10bf, maintainer_can_modify: true, mergeable state clean
  • Checks at this tip: 16 success + 1 cancelled (agent-audience, duplicate job) + 2 skipped; the only failure is enqueue-guard, which is red because this tip carries no PASS yet — expected for a T2
  • Open since 2026-11-17T06:17:18Z, i.e. 2h at the time of this comment; zero reviews so far

No code change and no new tip will follow from this comment.

Merged via the queue into ThinkFlowLab:main with commit 5c46977 Nov 17, 2026
21 of 22 checks passed
@Birfy
Birfy deleted the dev/1-social-failed-join branch November 17, 2026 08:19
@github-actions github-actions Bot removed the agent-pr Pull request opened by an agent label Nov 17, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

social: approve-all failure notice joins each person's reasons instead of running them together

2 participants