Skip to content

Commit 099a190

Browse files
committed
docs(devlog): record the 2026-08-31 non-priority-70 bug triage round
1 parent 6123be3 commit 099a190

6 files changed

Lines changed: 767 additions & 0 deletions

File tree

Lines changed: 103 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,103 @@
1+
# 260831 — bug triage round: everything the priority-70 train does not own
2+
3+
A concurrent session owns the >=70 train (`devlog/_plan/260831_prio70_train_round2/`):
4+
issues #3071, #3032, #3026, #3029, #3008, #3019 and their PRs #3069, #3056, #3040,
5+
#3020. This unit owns the rest of the open bug surface and drives it to zero, keeping
6+
only the handful that genuinely cannot be resolved from this tree.
7+
8+
## Frozen snapshot
9+
10+
Taken 2026-08-31T15:45:55Z against `dev` = `b4303bb9e`. Anything opened after that
11+
timestamp is queued for the next round and does not change this round's acceptance
12+
scope.
13+
14+
**Bug issues (11):** #3070 #3068 #3064 #3059 #3051 #3024 #3021 #2999 #2813 #1527 #1419
15+
**Bug-labelled PRs (13):** #3078 #3067 #3066 #3063 #3053 #3052 #3041 #3039 #3038 #3034
16+
#3003 #3000 #2989
17+
**Also in scope, not bug-labelled:** #3030 (`chore`), carried only as a wrong-branch
18+
janitorial closure. Audit round 1 caught this misclassification; see `002`.
19+
20+
Issue #3009 and PR #3039 are in scope. They are easy to confuse with the train's #3008
21+
and PR #3040 — different defect, different file, different lane.
22+
23+
## Why this roadmap is deliberately shallow
24+
25+
The prio-70 unit wrote six diff-level decade docs before implementing anything. That
26+
worked for six deep defects. This round has twenty-five items whose correct
27+
disposition is mostly *closure*, and pre-writing a diff for an item that turns out to
28+
be already fixed is wasted precision that then has to be un-written.
29+
30+
So this document locks only what a roadmap must lock: the scope, the cluster
31+
partition, the order, and the candidate disposition per item. **The diff-level design
32+
for each cluster is produced in that work-phase's own A phase**, against the tree as
33+
it stands when that phase starts, and recorded in the phase's own decade doc. This is
34+
an explicit, user-directed deviation from DIFFLEVEL-ROADMAP-01.
35+
36+
## Disposition vocabulary
37+
38+
Every item leaves this round through exactly one of:
39+
40+
| verdict | meaning |
41+
| --- | --- |
42+
| `MERGE` | the PR is correct and complete; squash after exact-head CI |
43+
| `CHERRY_PICK` | only part of the PR is correct; take those hunks |
44+
| `REIMPLEMENT` | the diagnosis is right and the remedy is wrong; rewrite on `dev` with a red-then-green regression |
45+
| `CLOSE_FIXED` | already fixed on `dev`; cite the commit |
46+
| `CLOSE_INVALID` | the claimed code path contradicts the tree |
47+
| `CLOSE_DUPLICATE` | name the survivor |
48+
| `CLOSE_NOT_REPRO` | no reproduction is possible against current `dev` |
49+
| `UNSOLVABLE` | stays open; name exactly what external input is missing |
50+
51+
A closure without a `file:line` or commit SHA in its comment does not count.
52+
53+
## Work-phase map
54+
55+
The order below is the audited order, not the original one. Audit round 1 found three
56+
file collisions the first ordering ignored, two of them with the concurrent >=70 train
57+
(`002`, findings 5-7). Train-blocked phases run late so an external dependency never
58+
stalls the round.
59+
60+
| # | wp | cluster | items | candidate disposition | blocked by |
61+
| --- | --- | --- | --- | --- | --- |
62+
| 0 | wp0 | this roadmap + live rescan | all of scope | — | — |
63+
| 1 | wp1 | closes with no code | #3068, PR #3030 | duplicate; wrong-branch | — |
64+
| 2 | wp2 | model catalog dated variants | #3024, PR #3034, PR #3041 | widen the suffix one-way; cherry-pick the merge tests | — |
65+
| 3 | wp3 | cursor discovery transport | #3051, PR #3052 | merge after rebase | — |
66+
| 4 | wp4 | windows service and scheduler | #3064 + PR #3067, #3009 + PR #3039 | one reimplement, one merge-after-fix | — |
67+
| 5 | wp7 | residual bug PRs | PR #3078, PR #3053 | reimplement on dev; merge | #3053 needs wp2 |
68+
| 6 | wp6 | account-pool auth and quota | PR #2989, then #2999 + PR #3000, then PR #3003 | merge; portable rewrite; merge | #2999 rewrite needs #2989 first (same file); #3003 needs train #3020 |
69+
| 7 | wp5 | upstream request and compact metadata | PR #3066, then PR #3063, then PR #3038 | merge; merge; close the duplicate | #3066 and #3063 share `openai-responses.ts`, so #3066 lands first and #3063 rebases onto it. Train #3089 merged at `a0d386b49`, so the external blocker is gone (`004`) |
70+
| 8 | wp9 | residual issue fixes | #3070, #1527, #3021, #3059 | four bounded reimplementations | — |
71+
| 9 | wp8 | closeout | — | receipts, residual set, final audit | all |
72+
73+
Each row is one full PABCD cycle. The candidate column is what this round's four
74+
read-only `xai/grok-4.6` lanes concluded; none of it is binding until that phase's own
75+
A phase confirms it against the tree as it stands then.
76+
77+
**Every phase re-reads `gh pr diff --name-only` for its own PRs before merging, and pairs
78+
that list against every other PR it is about to touch.** #3063 grew from two files to
79+
five during wp0 and picked up two train-owned files, which moved it from wp7 to wp5
80+
(`003`); pairing then exposed that it also collides with #3066 (`004`). A file list
81+
captured at scan time is not a fact about merge time, and neither is a blocker — train
82+
#3089 merged at `a0d386b49` while this roadmap was being audited.
83+
84+
## Declared unsolvable
85+
86+
#2813 (needs a live Luna Reserve account's `/v1/models` and `/api/models` dumps) and
87+
#1419 (needs macOS `.ips` crash frames on Bun 1.4.0, which the maintainer already
88+
declined to claim was fixed). Both are argued in `002`. Two of the allowed three-to-four
89+
slots are spent; the rest stay unspent until a phase earns one.
90+
91+
## Constraints this round runs under
92+
93+
- No local full suite. Focused `bun test tests/<file>.test.ts` or `bun run test:changed`
94+
only; whole-suite evidence comes from hosted exact-head CI or `ssh lidge`.
95+
- Every commit and push uses `--no-verify`. `dev` is protected, so every change lands
96+
through a branch and a PR, merged after exact-head CI is green.
97+
- Read-only `xai/grok-4.6` lanes, unlimited, for investigation and audit.
98+
- No file owned by the >=70 train is touched.
99+
100+
## Terminal outcome
101+
102+
`DONE` requires every scoped item terminal, at most four left open, each with a
103+
recorded reason. Receipts land in `070_outcome.md`.
Lines changed: 77 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,77 @@
1+
# 001 — live rescan verdicts for all 25 scoped items
2+
3+
Four read-only `xai/grok-4.6` high-effort lanes, split so no two lanes shared a
4+
verdict: (A) issues a prior scan believed the tree already answered, (B) the model
5+
catalog dated-variant cluster, (C) platform/service and cursor transport, (D) request
6+
metadata, account-pool auth, and the residual bug PRs. Every lane was instructed that
7+
the tree wins over the issue body and the PR description.
8+
9+
## The headline: the prior scan was wrong in both directions
10+
11+
The round-2 below-bar table (`260831_prio70_train_round2/000_plan.md`) was written to
12+
justify *exclusion* from a merge train, not to decide disposition. Read as disposition,
13+
it misclassifies five items:
14+
15+
| item | prior scan said | the tree says |
16+
| --- | --- | --- |
17+
| #3041 | reverse inference can resurrect retired ids | the author **removed** the reverse fold in `4e131140c`; the danger is in `aef4bec2`, which is no longer the head |
18+
| #3070 | model filtering landed in `b68edc077` | that commit is CLI/API only and does not touch `gui/src/pages/Logs.tsx`; the dashboard still cannot find a Terra row |
19+
| #1527 | all four named mechanisms are fixed on `dev` | all four SHAs are ancestors, but `envelope_exhausted` still silently full-replays for external-root models |
20+
| #3021 | one occurrence, no ciphertext captured | `structurallyValidFernetTokens` already exists, so a bounded output filter needs no reporter ciphertext |
21+
| #3053 | no linked user report | the runtime/catalog drift is real and the PR's tests drive production; absence of an issue number is not a defect |
22+
23+
#3059 and #1419 held up only halfway, and audit round 1 caught the other half. The tree
24+
does contradict #3059's unmount path, but a real focus residual survives that
25+
refutation, so it became a wp9 fix instead of a close. #1419 was reported against a Bun
26+
this tree no longer ships, but the maintainer explicitly declined to claim 1.4.0 fixed
27+
it, so it became the second declared `UNSOLVABLE` instead of a close. See `002`.
28+
29+
## Verdicts
30+
31+
| item | verdict | one-line basis | phase |
32+
| --- | --- | --- | --- |
33+
| #3068 | `CLOSE_DUPLICATE` | same author, body and `input[240]` log as #3071; author already said "superseded" | wp1 |
34+
| #3059 | `REIMPLEMENT` (was `CLOSE_INVALID`; see `002`) | the reported unmount cannot run — `refresh()` keeps stale data at `gui/src/client-resource.ts:339-341` — but the focus residual at `RestoreDialog.tsx:49-50` is real | wp9 |
35+
| #1419 | `UNSOLVABLE` (was `CLOSE_NOT_REPRO`; see `002`) | `27764f342` moved the pin to Bun 1.4.0 and 200 TLS-failure cases produced no SIGTRAP, but the maintainer explicitly declined to claim that fixed the reporter's trap | residual |
36+
| PR #3030 | `CLOSE_INVALID` | the branch is 61 files / +5114 of unrelated `main` work, and the classification it tests does not exist in `provider-routes.ts:957` | wp1 |
37+
| #3024 | `REIMPLEMENT` | widen the suffix matcher one-way only; a live base row is not callability evidence for a configured dated snapshot | wp2 |
38+
| PR #3034 | `MERGE_AFTER_REBASE` | the calendar matcher is the better vehicle; graft #3041's merge-loop tests, whose reverse test is the real resurrection guard | wp2 |
39+
| PR #3041 | `CHERRY_PICK` | take the two merge tests and the directional comment; leave `isDateSuffix`, which still folds `0231` | wp2 |
40+
| #3051 | via PR | — | wp3 |
41+
| PR #3052 | `MERGE_AFTER_REBASE` | one production line; a pre-header EOF is `status === 0` and must be `transport`, not `HTTP unknown` | wp3 |
42+
| #3064 | via PR | — | wp4 |
43+
| PR #3067 | `REIMPLEMENT` | `[^\\\\/]*` leaves a fully CJK segment with no anchors, so `...\\김병준\\...` matches `...\\Admin\\...`; restrict the lossy run to `[?\\uFFFD]*` and give `<UserId>` its own matcher | wp4 |
44+
| #3009 | via PR | — | wp4 |
45+
| PR #3039 | `MERGE_AFTER_REBASE` | correct remedy; restore `expect(probes).toBe(1)` and pin the 45s budget absolutely | wp4 |
46+
| PR #3066 | `MERGE_AFTER_REBASE` | strips at the noncanonical adapter boundary only, copy-on-write, ChatGPT preserved; tests drive `buildRequest` | wp5 |
47+
| PR #3038 | `CLOSE_DUPLICATE` | same defect, wrong layer (mutates canonical ChatGPT too) and its tests stay green with both call sites deleted | wp5 |
48+
| #2999 | `REIMPLEMENT` | refresh lock is keyed on `OPENCODEX_HOME` while the file lives in `CODEX_HOME`; coordinate on the existing native-main claim instead | wp6 |
49+
| PR #3000 | `REIMPLEMENT` (close in favor of the rewrite) | `dlopen(\"libc.so.6\")` breaks musl, and a late cancel discards an already-rotated grant | wp6 |
50+
| PR #3003 | `MERGE_AFTER_REBASE` | a failed WHAM prime writes no quota, so the account is stale forever; the PR's tests drive `primeCodexPoolQuotas` | wp6 |
51+
| PR #2989 | `MERGE_AFTER_REBASE` | the existing 503 test never re-enters, so it stays green on the broken path; the PR's tests do re-enter | wp6 |
52+
| PR #3078 | `REIMPLEMENT` on `dev` | both production hunks are right; it targets `main` and its test file does not typecheck | wp7 |
53+
| PR #3063 | `MERGE_AFTER_REBASE` | the second commit's tests do drive `handleResponsesCompact`; the "vacuous" reading was of the first commit | wp5 (moved: it now edits two train-owned files, `003`) |
54+
| PR #3053 | `MERGE_AS_IS` | already rebased onto `b4303bb9e`; mirrors `isModelTextOnly` at both catalog sites | wp7 |
55+
| #3070 | `REIMPLEMENT` | add a Logs model/provider query; the intercepted toggle stays Luna-only | wp9 |
56+
| #1527 | `REIMPLEMENT` | fail closed on `envelope_exhausted` for external-root models instead of silently full-replaying | wp9 |
57+
| #3021 | `REIMPLEMENT` | replace a client-visible Fernet payload with a structured error; do not widen recovery to `MESSAGE` | wp9 |
58+
| #2813 | `UNSOLVABLE` | needs `/v1/models` and `/api/models` dumps from an account actually in Reserve; a picker screenshot cannot separate proxy-missing from client-filter | residual |
59+
60+
## Residual candidates
61+
62+
Two are declared unsolvable after audit round 1: #2813 and #1419. The round budget
63+
allows three to four, so the remaining slots are held for phases that hit a genuine
64+
wall, not spent in advance.
65+
66+
## Note on phase numbering
67+
68+
wp9 (residual issue reimplementations: #3070, #1527, #3021, and #3059 after audit round
69+
1) was appended after this scan, because the roadmap assumed those would close without
70+
code. wp8 remains the closeout and runs last.
71+
72+
## This table is living
73+
74+
Two audit rounds moved four rows after they were first written (#3059, #1419, #3063,
75+
and #3030's label). PR file lists in particular are a moving target — #3063 grew from
76+
two files to five during wp0 — so every phase re-reads `gh pr diff --name-only` for its
77+
own PRs before merging rather than trusting this table's snapshot.
Lines changed: 101 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,101 @@
1+
# 002 — audit round 1: nine findings, seven upheld, one rebutted, one reclassified
2+
3+
One adversarial `xai/grok-4.6` round against `000` and `001`. Verdict FAIL. Every
4+
finding was re-checked against the tree by the main session before it was accepted or
5+
rebutted; the reviewer's own citations were not taken on trust either.
6+
7+
## Upheld — these change the plan
8+
9+
**1. #3059 is not a clean `CLOSE_INVALID`.** The lane's mechanism analysis is right:
10+
`refresh()` keeps stale data (`gui/src/client-resource.ts:339-341` — `shouldShowLoading`
11+
is true only when `data === undefined` or `forceLoading`), so the `if (!status)` branch
12+
at `gui/src/pages/integrations/FileIntegrationPage.tsx:175` is cold-load only and the
13+
reported unmount cannot run. But a real focus residual survives that refutation, and
14+
the code says so itself at `gui/src/pages/integrations/RestoreDialog.tsx:64-66`:
15+
16+
> The row's button is gone from the DOM in the collapsed case, so this is a best
17+
> effort: focus returns only if the trigger survived the close.
18+
19+
The reporter's diagnosis is wrong and their experience is real. Closing as invalid
20+
would discard the second half. **#3059 moves to wp9 as a bounded fix**: restore focus to
21+
a stable element when the trigger did not survive, rather than dropping focus to
22+
`<body>`. The trigger is the per-row button in
23+
`gui/src/pages/integrations/RollbackHistory.tsx:48-56`, which is exactly the element the
24+
collapsed case removes. The comment quoted above is at `RestoreDialog.tsx:49-50`, inside
25+
the effect cleanup — not at `:64-66`, which is the `submit` body.
26+
27+
**2. PR #3030 is `chore`, not `bug`.** `gh pr view 3030 --json labels` returns
28+
`["chore","intake: hygiene-blocked"]`. The frozen scope called it a bug PR. Corrected
29+
count: **13 bug-labelled PRs** (excluding the train's #3020) plus #3030, which stays in
30+
scope only as a wrong-branch janitorial closure and is labelled as such. The citation
31+
`provider-routes.ts:957` was also imprecise — line 957 is the `jsonResponse` inside the
32+
catch; the point is that the whole catch block (`:955-965`) has no timeout
33+
classification and `rg "Connection test timed out" src tests` returns nothing.
34+
35+
**3. #1419 must not be closed.** The maintainer's own last comment keeps it open in
36+
writing: "That is encouraging but **not** proof your crash is fixed... Claiming 1.4
37+
resolved your specific trap would go beyond what I can show." Closing it as not-repro
38+
would contradict a recorded maintainer position. **#1419 becomes the second
39+
`UNSOLVABLE`**: it needs macOS `DiagnosticReports` `.ips` frames from a recurrence on
40+
Bun 1.4.0, which no one on this tree can synthesize.
41+
42+
**5. PR #3066 collides with the train.** #3066 and the train's #3089 (the reopened
43+
#3071 fix, head `codex/3071-web-search-query`) both edit
44+
`src/adapters/openai-responses.ts`, and #3089 rewrites `backfillWebSearchQueries`
45+
immediately above #3066's insertion point. **Ordering constraint: wp5 does not merge
46+
until #3089 lands, then rebases onto that head.** If #3089 has not landed when wp5 comes
47+
up, wp5 waits and a later phase runs first.
48+
49+
**6. PR #3003 collides with the train.** #3003 and the train's #3020 both edit
50+
`src/codex/auth-api.ts`, and both rewrite `primeCodexPoolQuotas` /
51+
`fetchPoolAccountQuota`. **Ordering constraint: #3020 lands first, then #3003 rebases.**
52+
53+
**7. Internal collision inside this round.** wp2 (#3034/#3041) and wp7 (#3053) both edit
54+
`src/codex/catalog/provider-fetch.ts`. They are not disjoint. **wp2 lands before #3053.**
55+
56+
**8. The phase map contradicted the verdict table.** `000` put #3070 and #1527 in wp1 as
57+
closures while `001` marked both `REIMPLEMENT`; #3021 had the same split. Executing wp1
58+
from `000` would have closed two issues this scan had just proved still need code. The
59+
`000` table is corrected and wp9 is now scheduled in it.
60+
61+
**9. #3068 closes only as a duplicate of #3071.** The survivor is open and owned by the
62+
other train, so the closing comment names #3071 and #3089 and claims nothing about a
63+
fix being present.
64+
65+
## Rebutted
66+
67+
**4. #3041's `isDateSuffix` does accept `0231`.** The reviewer read the rejection tests
68+
(`0001`, `1300`, `1240`) and concluded February 31 is rejected too. It is not:
69+
70+
```
71+
$ git show refs/tmp/pr-3041:src/codex/catalog/provider-fetch.ts | rg -A6 'function isDateSuffix'
72+
948:function isDateSuffix(suffix: string): boolean {
73+
949: if (/^\d{8}$/.test(suffix)) return true;
74+
950: if (!/^\d{4}$/.test(suffix)) return false;
75+
951: const month = Number(suffix.slice(0, 2));
76+
952: const day = Number(suffix.slice(2));
77+
953: return month >= 1 && month <= 12 && day >= 1 && day <= 31;
78+
954:}
79+
```
80+
81+
`0231` is month 2, day 31: both bounds pass, so it folds. `1240` is rejected because day
82+
40 exceeds 31, which is what the reviewer's cited test actually proves. The eight-digit
83+
branch is worse — bare `/^\d{8}$/` folds `20250229`. #3034's calendar matcher rejects
84+
both. The `CHERRY_PICK` verdict stands unchanged.
85+
86+
## Revised residual set
87+
88+
| item | why it cannot be resolved this round |
89+
| --- | --- |
90+
| #2813 | needs `/v1/models` and `/api/models` dumps from an account actually in Luna Reserve; a picker screenshot cannot separate proxy-missing from client-filter, and one blind catalog-field PR (#2862) already failed |
91+
| #1419 | needs macOS `.ips` crash frames from a recurrence on the Bun this tree ships; the maintainer already declined to claim 1.4.0 fixed it |
92+
93+
Two of the allowed three-to-four slots are spent. The rest are held for phases that hit
94+
a real wall.
95+
96+
## Corrected phase order
97+
98+
**Superseded by `003` and `004`.** The order this round produced put #3063 in wp7 and left
99+
#2989 and #3000 unordered against each other; rounds 2 and 3 fixed both. `000` carries
100+
the authoritative order — this section is kept only so the amendment history reads in
101+
sequence.

0 commit comments

Comments
 (0)