-
Notifications
You must be signed in to change notification settings - Fork 1.3k
docs(devlog): close the dev hardening unit with a promotion-readiness statement #2752
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Changes from all commits
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,60 @@ | ||
| # wp7 — promotion readiness at `2a72cc017` | ||
|
|
||
| This is the closing record for the 260827 dev hardening unit. It states what is proven | ||
| about the current `dev` head, what is deliberately not on it, and what a promoter still | ||
| has to do. It is not an approval to promote. | ||
|
|
||
| ## The head | ||
|
|
||
| `dev` = `2a72cc0173af36d4b8172b70ba0a3384db9e6047`, 283 commits ahead of `main` | ||
| (`ec51e42d7`, v2.33.0). `package.json` reads `2.34.0`, so the version line is ahead of the | ||
| published channel rather than behind it — which is the whole subject of wp2 and is now | ||
| guarded by `tests/release-version-line.test.ts`. | ||
|
|
||
| ## What this unit changed | ||
|
|
||
| | Work phase | PR | Merged as | What it closed | | ||
| | --- | --- | --- | --- | | ||
| | wp1 | #2738 | `5dfee1a05` | inventory and remediation roadmap | | ||
| | wp2 | #2739 | `ca3b379e1` | dev carried a version behind its own releases | | ||
| | wp2b | #2743 | `a57b9620a` | the version guard could go inert without `fetch-tags` | | ||
| | wp3 | #2745 | **not merged** | credential identity on OAuth 429 rotation | | ||
| | wp4 | #2746 | `802f04adc` | locale pages that contradicted the code | | ||
| | wp5 | #2749 | `5000321e6` | `doctor:gui` failed on dev, so prepush was routinely bypassed | | ||
| | wp6 | #2751 | `2a72cc017` | two invariants AGENTS.md claimed were enforced, and were not | | ||
|
|
||
| ## Proven at this head | ||
|
|
||
| - Full `bun run test` on the Linux host: rc=0. The three runs during wp6 reported 15,324 | ||
| passing tests and 0 failures with no `(fail)` lines; the wp5 runs reported 15,318. | ||
| - Cross-platform CI green on each merged head, covering Linux, Windows, and macOS. The | ||
| macOS leg is the slow one at roughly 9-11 minutes and is always last to report. | ||
| - `bun x tsc --noEmit`, `bun run lint:gui`, and `bun run doctor:gui` all exit 0 locally on | ||
| `dev` — the last of those for the first time this unit, which is what wp5 was for. | ||
|
|
||
| ## Not on this head, on purpose | ||
|
|
||
| **PR #2745 (wp3) is open and must stay open.** It changes credential identity handling on | ||
| OAuth 429 rotation, which `MAINTAINERS.md` puts behind explicit security review. It was not | ||
| self-merged and `maintainer-sponsored` was deliberately not applied. Anyone promoting `dev` | ||
| should know that the failover identity drift described in | ||
| `devlog/_plan/260827_dev_hardening/020_wp3_failover_identity.md` is still present on `dev`. | ||
|
|
||
| The review lane that examined it refuted the exploit's reachability — the missing | ||
| `continue recovery` acts as an accidental guard — so this is a correctness defect rather | ||
| than a live vulnerability. That is the reason it is a normal review queue item and not a | ||
| release blocker. | ||
|
Comment on lines
+43
to
+46
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more.
Because PR #2745 remains open and the credential defect is still present, these lines publish a severity assessment and the accidental guard that currently prevents exploitation in a tracked public directory. Repository policy requires findings, bypass reasoning, and assessments for unfixed security defects to remain in scratch space; remove these details from the promotion record until the fix has shipped. AGENTS.md reference: AGENTS.md:L103-L108 Useful? React with 👍 / 👎. |
||
|
|
||
| ## What a promoter still has to do | ||
|
|
||
| Nothing here substitutes for the release procedure in `MAINTAINERS.md`. Promotion is | ||
| maintainer-controlled, and `scripts/release.ts` remains the release authority. This record | ||
| only establishes that the integration branch is in a state worth promoting from. | ||
|
|
||
| ## What this unit did not attempt | ||
|
|
||
| The audit walked the core/Lab boundary, the startup activation window, repository hygiene, | ||
| the version line, locale docs, and the local gates. It did not audit provider adapters, | ||
| the GUI beyond its lint configuration, or the release workflow itself. A green suite at | ||
| this head is evidence about the code that has tests, which is the ordinary limit of this | ||
| kind of statement and worth saying out loud rather than implying otherwise. | ||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Calling this the unit's closing record omits the already-tracked wp8 result:
060_wp8_launcher_flake.mdrecords two macOS failures oftests/update-stop-first.test.tsand explicitly classifies the recovery slowness as a known open defect. Consequently this closeout can tell a promoter the branch is ready while an unresolved readiness failure remains in the same unit; retain an open status and include wp8's disposition, or resolve it and then move the completed unit to_fin.AGENTS.md reference: AGENTS.md:L83-L86
Useful? React with 👍 / 👎.