[tracking202]

Every 1.9.79 PR added its release-note lines at the end of the version's list, so every merge of two of them conflicted there. Resolving by keeping both sides kept two things:

• a line both sides already had;
• the older wording of a line one side had reworded.

Under Version 1.9.79, changelogs.txt held 124 entries for 81 changes:

• 35 exact repeats. PayKickstart's IPN was listed four times. JVZoo, ClickBank's INS and the LTV webhooks were each listed three times; others twice.
• 8 older wordings kept beside the newer. The newer is the one each change's merged PR head carries, checked against the heads of 1.9.79: PayKickstart's product IPN, verified with each campaign's key #268, 1.9.79: JVZoo's JVZIPN at jvzoo.php, for affiliates and vendors (stacked on #268) #272, 1.9.79: ClickBank's INS works for every account, decrypted with each account's own key #276, 1.9.79: ClickBank rebills are recorded, and refunds and chargebacks take a sale back #288, 1.9.79: offline conversion files for Google Ads and Microsoft Advertising #293, 1.9.79: P202_SERVER_STATE_DIR keeps a directory per install, so a reinstall under it replays nothing #317 and 1.9.79: sign-out and switching a user off revoke remember-me keys #324. The rewordings it keeps:
  • a deleted account's stored keys verify nothing (PayKickstart, JVZoo, ClickBank);
  • a second ClickBank refund is refused;
  • Google's one-conversion-per-second rule in the offline files;
  • P202_SERVER_STATE_DIR per install;
  • switching a user off revoking their remember-me keys.

Each change now has one entry, at the place its first copy held, in the wording its PR ended with. The removed set is exactly those 43 copies (checked by script): no entry is added or reworded, and nothing outside 1.9.79 changes. 202-config/changelogs.json had no repeats and is untouched.

The check

tests/Release/ChangelogEntriesAreUniqueTest:

• A repeated entry is refused in any version of either file. Every released version passes.
• For the version version.php names, two entries that share 60% of their five-word runs are refused. On 1.9.79 every reworded copy scored 0.76 to 0.97, and the closest two distinct changes scored 0.18.
• It is not asked of released versions: 1.9.56 lists one change as both an Update and a Fix (0.78), and it shipped that way.

Plants:

• With 1.9.79's merged changelogs.txt put back, both tests fail.
• With only the eight reworded copies added back to the cleaned file, the wording test alone fails, naming all nine pairs.
• Both were restored with cmp, and the suite was re-run green.

The merge tool that produced most of these was the coordinator's changelog resolver. It kept the incoming side's copy of a line this branch had reworded. It now drops it, in both the text and the JSON resolver.

No backlog item: K2SP's agent found it on 1.9.79 after the merges, and I counted it.

Verification

Run in a worktree at 8288e8b (origin/1.9.79 c6d5167 plus this commit), PHP 8.3.

.claude/skills/p202-verify/scripts/verify.sh --changed, run before the four long lines in the new test were wrapped:

Scope report
------------
  patterns   PASS
  phpcs      FAIL
  phpstan    FAIL  phpstan.neon.dist failed, phpstan-strict.neon passed
  syntax     PASS
  unit       PASS

Not covered by this script:
  live      stand up an instance, drive the path a user would take
  agent     add and run a case under tests/fixtures/agent-eval/cases/
  ci        a workflow still in_progress has not passed; read the run

• phpcs: the four lines over 120 characters in the new test. They were wrapped, and phpcs --standard=PSR12 on the file is now clean.
• phpstan: 1.9.79's own five errors, in account.php, offline_conversions.php and ppc_accounts.php, red on c6d5167 alone. 1.9.79: read the three request values 1.9.79's own merges left cast (PHPStan red on 1.9.79) #328 fixes them. PHPStan on the new test file reports nothing.
• unit: 7,576 tests, 5 skipped. The new test is 3 tests, 161 assertions.

There is no live or agent path: this is the release-notes text and a test.

---

Generated by Claude Code

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	⚠️ Failed 2026-10-28T16:26:16.792319Z We couldn’t complete this review. Comment @codex review to try again.	8288e8b	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

[tracking202]

What is red here, and why it is not this PR's:

• PHPStan (custom rules) is red on 1.9.79's own tip (c6d5167): three request casts left by its merges. 1.9.79: read the three request values 1.9.79's own merges left cast (PHPStan red on 1.9.79) #328 fixes them. 580bd44 merges 1.9.79: read the three request values 1.9.79's own merges left cast (PHPStan red on 1.9.79) #328's branch here (PHPStan locally: 0 errors; ChangelogEntriesAreUniqueTest still passes with 1.9.79: read the three request values 1.9.79's own merges left cast (PHPStan red on 1.9.79) #328's changelog line), and the merge adds nothing once 1.9.79: read the three request values 1.9.79's own merges left cast (PHPStan red on 1.9.79) #328 is in 1.9.79.
• claude-review and adversarial-review did not run. The Claude app runs a review only when its workflow file matches master's copy. 6834377 (1.9.79: an unfinished review fails its check, and both reviews read the whole diff #275) changed both review workflows on 1.9.79, so neither runs on any PR into 1.9.79 until master carries those files. No PR into 1.9.79 can fix that, and re-running gives the same refusal.

---

Generated by Claude Code