Skip to content

Tooling batch: nine retro tickets - #433

Merged
Kyu-seok merged 31 commits into
masterfrom
integration/tickets-2026-11-20
Nov 22, 2026
Merged

Kyu-seok merged 31 commits into
masterfrom
integration/tickets-2026-11-20

Conversation

@Kyu-seok

@Kyu-seok Kyu-seok commented Nov 20, 2026 •

Copy link
Copy Markdown
Owner

Nine tickets with no spec, landed on one integration branch.

State directory: ~/.cache/implement-spec/tickets-2026-11-20/ (brief: BRIEF.md, merge log: merge-log.md). Integration tip: cea22201.

Merged

#432

Cause. open() wrote each pass's language and theme into localStorage on the home page, where the previous pass's signed-in app was still running. When that app's GET /auth/me answered after the write (slow backend) and before the problem page replaced it (slow Vite), it wrote the user's locale (ko) back. The signed-out English pass then waited for "Description" on a Korean page until its waits ran out, and atTab read a tab that wasn't there. CPU throttling alone (20x, 40x through CDP) does not reproduce it.

Reproduction on master (branch servers up; holds /api/auth/me 2 s and the problem page 5 s in Chrome):

cd <worktree>
perl -pe 's/(await send\(.Page\.enable.\); await send\(.Runtime\.enable.\);)/$1 await send("Fetch.enable", { patterns: [{ urlPattern: "*\/api\/auth\/me*" }, { urlPattern: "*\/problems\/*", resourceType: "Document" }] }); listeners.push(m => { if (m.method === "Fetch.requestPaused") setTimeout(() => send("Fetch.continueRequest", { requestId: m.params.requestId }).catch(() => {}), m.params.request.url.includes("\/auth\/me") ? 2000 : 5000); });/' \
  scripts/walkthrough-app-check.mjs > scripts/.repro-432.mjs
node scripts/.repro-432.mjs <vite url> <api url> <out dir> find-menu-price; rm scripts/.repro-432.mjs

On master:

FAIL out-390-light-en: tabs in en (null)
FAIL out-390-light-en: Walkthrough tab shows
TypeError: Cannot read properties of undefined (reading 'right')
    at atTab (…)

exit 1, about 58 s. On this branch the same command exits 0, because the fix removes the race: open() now sets localStorage in the new document before the app runs.

A wait cut short, on this branch: WALKTHROUGH_WAIT_MS=20 node scripts/walkthrough-app-check.mjs <vite> <api> <out> <slug> prints FAIL out-390-light-en: tabs did not show within 0.02 s with out-390-light-en-stopped.png for each pass, exit 1, no stack trace. The named line was not captured under the reproduction itself, which no longer fails.

Under load: three runs in a row during scripts/gradle-test.sh :backend:test --tests 'com.codebite.problem.*' all exit 0 (41 s, 27 s, 33 s); the slowest first load was 4.2 s, so the 15 s bound stays.

#405

Every fact of the old "Parallel sessions" paragraph, against its place in docs/agents/parallel-sessions.md (wording carried over unchanged):

Old fact Where it is now
Several sessions often run here at once Opening line (and the CLAUDE.md pointer)
Leave the main checkout on master; all branch work in your own worktree Worktrees, first bullet (and the pointer)
A git switch or commit in the shared checkout moves every other session's files Worktrees, first bullet
git worktree add <scratchpad>/wt -b <branch> origin/master Worktrees, second bullet
npm ci in the worktree's frontend/ before any frontend check: no node_modules, the main checkout's may be stale Worktrees, second bullet
The CLAUDE.md a session starts with is the main checkout's and can trail origin/master; so can a skill loaded at session start What trails origin/master, first bullet
After git worktree add, run git -C <worktree> diff $(git rev-parse HEAD) -- CLAUDE.md .claude/skills from the main checkout and read what landed (a newer script, a skill step) What trails origin/master, first bullet
Read and edit source only under the worktree path; a patch written from the main checkout's text misses What trails origin/master, second bullet
8 GB, so one Gradle run at a time through scripts/gradle-test.sh <gradle args>, which waits for another worktree's run; hooks and scripts use it too Gradle and memory, first bullet (the rule is also in the pointer)
A parallel implementer runs the test package it touches (--tests 'com.codebite.<package>.*'); the full suite is the merge gate's Gradle and memory, second bullet
Bash grep is the claude binary's ugrep, killed (exit 137) when memory runs low with whatever pipes into it (migrate | grep left a migration unapplied); write long output to a file, use /usr/bin/grep in pipelines Gradle and memory, third bullet
The pre-commit hook refuses commits on master; the commit-msg hook a subject over 72 characters Commits
scripts/ship-branch.sh <branch>… from the main checkout: stops on unticked - [ ] unless --allow-unticked, waits for CI, squash-merges, deletes the branch (remote, local, worktree, database) Merging, first bullet (the sentence as #431 left it)
A project hook refuses gh pr merge --delete-branch, which deletes live worktrees Merging, second bullet
scripts/merge-ticket.sh <branch> <title> [...] from the integration worktree: merges in order, runs the checks a clean merge skips, pushes only when green (header has details); an orchestrator runs it directly, a subagent only for a conflict Merging, third bullet
(new, from the status update) the compose containers are shared; simulate a stopped one with a fake docker on PATH or DOCKER_HOST Shared containers

#403

#403: why every deploy rolled the backend and recreated nginx

Root cause. The server runs Docker 29.3.0 with the containerd image store and compose 5.1.0. There an image's id is the digest of an index ("manifest list") holding the image manifest and BuildKit's provenance attestation, and the attestation is new on every build, fully cached or not. provenance: false in the compose file (#347) was meant to stop it, but compose 5.1.0 leaves a false out of the bake definition it hands buildx (docker compose build --print shows an attest entry for true or mode=max, none for false), so buildx attached its default attestation anyway.

The deploy logs say the same: in all 12 deploys read (44687133950 to 44734872864) every build step is CACHED and the exporting manifest and exporting config digests repeat from run to run, while exporting attestation manifest and exporting manifest list are new each time. All 12 rolled the backend and recreated codebite-nginx and codebite-worker-1. The two guesses on record (a missed build cache, a jar that isn't reproducible) are both ruled out by those lines.

Loop (an isolated docker:29.3.0-dind daemon, which ships compose 5.1.0: the server's versions; this machine's Docker 20.10 has neither attestations nor the containerd store, so it cannot show the bug):

  1. docker compose -f infra/docker-compose.prod.yml build backend twice from the same tree, comparing docker image inspect -f '{{.Id}}' codebite-backend.
    • Before: sha256:273d60a7… then sha256:ad1e6119…, red. The second build exported the same config and a new attestation manifest.
    • With BUILDX_NO_DEFAULT_ATTESTATIONS=1: sha256:07b464f2… both times, green.
  2. Build, up -d, build, up -d on a service shaped like nginx (built image, fixed container name).
    • Before: the second up prints Recreate, red. After: Running, green.

Fix. scripts/deploy-prod.sh exports BUILDX_NO_DEFAULT_ATTESTATIONS=1 for every build it starts (the rollback's up --build too); the three dead provenance: false keys are removed. scripts/deploy-prod.test.sh checks a compose call from the script carries the variable.

nginx. The recreate had the same cause: its image id changed on every build, so up -d replaced it (about 2.1 s without nginx in run 44687133950: Recreate 06:22:16.07, Started 06:22:18.15). With stable ids a deploy that doesn't change frontend/ leaves nginx and the worker running. A deploy that does change the frontend still recreates nginx and fails requests for about 2 s.

Owed after merge.

  • The deploy of this change itself still rolls once and recreates nginx and the worker once: the ids change form, from an index digest to a manifest digest.
  • The next merge that touches nothing under backend/, common/, gradle/ or the root Gradle files should log --- Backend image and config unchanged: keeping the running replicas ---, with Container codebite-nginx Running instead of Recreate. Post that run id on Deploy: a no-backend merge still rolls the backend, and nginx recreate returns 502 #403.

Not tested. Whether the jar is byte-identical when Gradle really reruns (BuildKit layer cache pruned): every build seen was fully cached.

#388

Walkthroughs condensed (#388)

Walkthrough Longest run before after Other runs before -> after How
n-queens 118 33 12 -> 12 in full until an arrangement is found and a dead end backed out of (frames 1-23: two arrangements, then the first dead end), then one frame per further arrangement and one for the search left after the last
list-bracket-pairings 92 30 12 -> 12 the first two strings in full, then one frame per further string
restore-wifi-ip 50 44 7 -> 7, 26 -> 25 each pruned branch (a segment with no address below it) is one frame
merge-cafe-queues 38 27 25 -> 20, 29 -> 20 heapify, and each heappop or heappush with its sift, is one frame

Every run's last frame still carries the sample's expected output (the recorder refuses a run that doesn't, and WalkthroughsMatchSamplesTest passes unchanged). No other Walkthrough's JSON changed.

restore-wifi-ip only drops from 50 to 44: the lengths a call passes over were already folded into captions, so its length is the nine addresses' calls and returns, not pruned branches. Shortening it further needs another decision (for example the first two addresses in full, then one frame per address, as the other two backtracking ones).

Folding-frame screenshots (branch app, headless Chrome, dark)

n-queens, 1440 and 390 wide, English:

Korean: 1440, 390

list-bracket-pairings, 1440 and 390 wide, English:

Korean: 1440, 390

restore-wifi-ip, 1440 and 390 wide, English:

Korean: 1440, 390

merge-cafe-queues, 1440 and 390 wide, English:

Korean: 1440, 390

Every folding frame's caption, English and Korean

n-queens: folding frames of the longest run (n = 5)

  • Frame 24
    • EN: Skipping ahead, the search goes on the same way: after 2 more queens lifted and 5 placed, every row has a queen again, so columns [1,3,0,2,4] are the 3rd arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 2번 들어 올리고 5번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [1,3,0,2,4] 배치가 3번째로 답에 들어갑니다.
  • Frame 25
    • EN: Skipping ahead, the search goes on the same way: after 6 more queens lifted and 6 placed, every row has a queen again, so columns [1,4,2,0,3] are the 4th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 6번 들어 올리고 6번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [1,4,2,0,3] 배치가 4번째로 답에 들어갑니다.
  • Frame 26
    • EN: Skipping ahead, the search goes on the same way: after 5 more queens lifted and 5 placed, every row has a queen again, so columns [2,0,3,1,4] are the 5th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 5번 들어 올리고 5번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [2,0,3,1,4] 배치가 5번째로 답에 들어갑니다.
  • Frame 27
    • EN: Skipping ahead, the search goes on the same way: after 4 more queens lifted and 4 placed, every row has a queen again, so columns [2,4,1,3,0] are the 6th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 4번 들어 올리고 4번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [2,4,1,3,0] 배치가 6번째로 답에 들어갑니다.
  • Frame 28
    • EN: Skipping ahead, the search goes on the same way: after 5 more queens lifted and 5 placed, every row has a queen again, so columns [3,0,2,4,1] are the 7th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 5번 들어 올리고 5번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [3,0,2,4,1] 배치가 7번째로 답에 들어갑니다.
  • Frame 29
    • EN: Skipping ahead, the search goes on the same way: after 6 more queens lifted and 6 placed, every row has a queen again, so columns [3,1,4,2,0] are the 8th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 6번 들어 올리고 6번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [3,1,4,2,0] 배치가 8번째로 답에 들어갑니다.
  • Frame 30
    • EN: Skipping ahead, the search goes on the same way: after 7 more queens lifted and 7 placed, every row has a queen again, so columns [4,1,3,0,2] are the 9th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 7번 들어 올리고 7번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [4,1,3,0,2] 배치가 9번째로 답에 들어갑니다.
  • Frame 31
    • EN: Skipping ahead, the search goes on the same way: after 4 more queens lifted and 4 placed, every row has a queen again, so columns [4,2,0,3,1] are the 10th arrangement.
    • KO: 같은 방식으로 이어지는 탐색을 건너뜁니다. 퀸을 4번 들어 올리고 4번 놓은 뒤 다시 모든 행에 퀸이 놓였으니, 열 [4,2,0,3,1] 배치가 10번째로 답에 들어갑니다.
  • Frame 32
    • EN: Skipping the rest of the search: no column is left to try, so the calls return one by one, lifting the last 5 queens, and the board is empty again.
    • KO: 나머지 탐색을 건너뜁니다. 더 시도할 열이 없어 호출이 차례로 돌아가며 남은 퀸 5개를 들어 올리고, 판은 다시 비었습니다.

list-bracket-pairings: folding frames of the longest run (pairCount = 4)

  • Frame 18
    • EN: Skipping 5 steps like the ones before: build(opened=3, closed=1) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ((())()) is the 3rd string.
    • KO: 앞과 같은 5단계를 건너뜁니다. build(opened=3, closed=1) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ((())()) 문자열이 3번째로 목록에 들어갑니다.
  • Frame 19
    • EN: Skipping 4 steps like the ones before: build(opened=3, closed=2) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ((()))() is the 4th string.
    • KO: 앞과 같은 4단계를 건너뜁니다. build(opened=3, closed=2) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ((()))() 문자열이 4번째로 목록에 들어갑니다.
  • Frame 20
    • EN: Skipping 7 steps like the ones before: build(opened=2, closed=0) is back from its ( branch and tries ), and the calls after it fill the prefix again, so (()(())) is the 5th string.
    • KO: 앞과 같은 7단계를 건너뜁니다. build(opened=2, closed=0) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, (()(())) 문자열이 5번째로 목록에 들어갑니다.
  • Frame 21
    • EN: Skipping 5 steps like the ones before: build(opened=3, closed=1) is back from its ( branch and tries ), and the calls after it fill the prefix again, so (()()()) is the 6th string.
    • KO: 앞과 같은 5단계를 건너뜁니다. build(opened=3, closed=1) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, (()()()) 문자열이 6번째로 목록에 들어갑니다.
  • Frame 22
    • EN: Skipping 4 steps like the ones before: build(opened=3, closed=2) is back from its ( branch and tries ), and the calls after it fill the prefix again, so (()())() is the 7th string.
    • KO: 앞과 같은 4단계를 건너뜁니다. build(opened=3, closed=2) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, (()())() 문자열이 7번째로 목록에 들어갑니다.
  • Frame 23
    • EN: Skipping 6 steps like the ones before: build(opened=2, closed=1) is back from its ( branch and tries ), and the calls after it fill the prefix again, so (())(()) is the 8th string.
    • KO: 앞과 같은 6단계를 건너뜁니다. build(opened=2, closed=1) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, (())(()) 문자열이 8번째로 목록에 들어갑니다.
  • Frame 24
    • EN: Skipping 4 steps like the ones before: build(opened=3, closed=2) is back from its ( branch and tries ), and the calls after it fill the prefix again, so (())()() is the 9th string.
    • KO: 앞과 같은 4단계를 건너뜁니다. build(opened=3, closed=2) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, (())()() 문자열이 9번째로 목록에 들어갑니다.
  • Frame 25
    • EN: Skipping 8 steps like the ones before: build(opened=1, closed=0) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ()((())) is the 10th string.
    • KO: 앞과 같은 8단계를 건너뜁니다. build(opened=1, closed=0) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ()((())) 문자열이 10번째로 목록에 들어갑니다.
  • Frame 26
    • EN: Skipping 5 steps like the ones before: build(opened=3, closed=1) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ()(()()) is the 11th string.
    • KO: 앞과 같은 5단계를 건너뜁니다. build(opened=3, closed=1) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ()(()()) 문자열이 11번째로 목록에 들어갑니다.
  • Frame 27
    • EN: Skipping 4 steps like the ones before: build(opened=3, closed=2) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ()(())() is the 12th string.
    • KO: 앞과 같은 4단계를 건너뜁니다. build(opened=3, closed=2) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ()(())() 문자열이 12번째로 목록에 들어갑니다.
  • Frame 28
    • EN: Skipping 6 steps like the ones before: build(opened=2, closed=1) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ()()(()) is the 13th string.
    • KO: 앞과 같은 6단계를 건너뜁니다. build(opened=2, closed=1) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ()()(()) 문자열이 13번째로 목록에 들어갑니다.
  • Frame 29
    • EN: Skipping 4 steps like the ones before: build(opened=3, closed=2) is back from its ( branch and tries ), and the calls after it fill the prefix again, so ()()()() is the 14th string.
    • KO: 앞과 같은 4단계를 건너뜁니다. build(opened=3, closed=2) 호출이 ( 쪽에서 돌아와 ) 괄호를 시도하고 이어지는 호출이 접두사를 다시 채워, ()()()() 문자열이 14번째로 목록에 들어갑니다.

restore-wifi-ip: folding frames of the longest run (digits = "19216811")

  • Frame 3
    • EN: "9" as segment 2 is a dead end: the 2 calls below it, not drawn, try "216" and reach no address, so "9" comes back off.
    • KO: "9" 조각을 2번째 조각으로 놓으면 그 아래 호출 2개(그리지 않음)에서 "216" 조각을 시도해도 주소가 나오지 않는 막다른 길이므로, 다시 뗍니다.
  • Frame 5
    • EN: "1" leaves 4 digits, too many for 1 segment; "16" as segment 3 is a dead end: the 1 call below it, place(start=5), not drawn, tries 3 lengths and finds no segment 4 ("8" would leave 2 digits after segment 4, "81" would leave 1 digit after segment 4, "811" is over 255), so "16" comes straight back off.
    • KO: "1" 조각 뒤로 숫자 4개가 남아 조각 1개에 너무 많음, "16" 조각을 3번째 조각으로 놓으면 그 아래 place(start=5) 호출 1개(그리지 않음)에서 길이 3가지를 시도해도 4번째 조각을 찾지 못하는("8" 조각은 4번째 조각 뒤로 숫자 2개를 남김, "81" 조각은 4번째 조각 뒤로 숫자 1개를 남김, "811" 조각은 255 초과) 막다른 길이므로, 바로 다시 뗍니다.
  • Frame 11
    • EN: "1" leaves 4 digits, too many for 1 segment; "16" as segment 3 is a dead end: the 1 call below it, place(start=5), not drawn, tries 3 lengths and finds no segment 4 ("8" would leave 2 digits after segment 4, "81" would leave 1 digit after segment 4, "811" is over 255), so "16" comes straight back off.
    • KO: "1" 조각 뒤로 숫자 4개가 남아 조각 1개에 너무 많음, "16" 조각을 3번째 조각으로 놓으면 그 아래 place(start=5) 호출 1개(그리지 않음)에서 길이 3가지를 시도해도 4번째 조각을 찾지 못하는("8" 조각은 4번째 조각 뒤로 숫자 2개를 남김, "81" 조각은 4번째 조각 뒤로 숫자 1개를 남김, "811" 조각은 255 초과) 막다른 길이므로, 바로 다시 뗍니다.
  • Frame 16
    • EN: "6" as segment 3 is a dead end: the 1 call below it, place(start=5), not drawn, tries 3 lengths and finds no segment 4 ("8" would leave 2 digits after segment 4, "81" would leave 1 digit after segment 4, "811" is over 255), so "6" comes straight back off.
    • KO: "6" 조각을 3번째 조각으로 놓으면 그 아래 place(start=5) 호출 1개(그리지 않음)에서 길이 3가지를 시도해도 4번째 조각을 찾지 못하는("8" 조각은 4번째 조각 뒤로 숫자 2개를 남김, "81" 조각은 4번째 조각 뒤로 숫자 1개를 남김, "811" 조각은 255 초과) 막다른 길이므로, 바로 다시 뗍니다.
  • Frame 29
    • EN: "6" as segment 3 is a dead end: the 1 call below it, place(start=5), not drawn, tries 3 lengths and finds no segment 4 ("8" would leave 2 digits after segment 4, "81" would leave 1 digit after segment 4, "811" is over 255), so "6" comes straight back off.
    • KO: "6" 조각을 3번째 조각으로 놓으면 그 아래 place(start=5) 호출 1개(그리지 않음)에서 길이 3가지를 시도해도 4번째 조각을 찾지 못하는("8" 조각은 4번째 조각 뒤로 숫자 2개를 남김, "81" 조각은 4번째 조각 뒤로 숫자 1개를 남김, "811" 조각은 255 초과) 막다른 길이므로, 바로 다시 뗍니다.

merge-cafe-queues: folding frames of the longest run (queues = [[2,5,8],[1,3,7],[4,6]])

  • Frame 3
    • EN: heapify sifts each parent down, drawn as one step: after 1 swap the smallest front, 1 from q1, is on top.
    • KO: heapify가 부모를 차례로 아래로 내리는 과정을 한 단계로 그립니다. 자리 바꿈 1번 끝에 가장 작은 앞 노드인 q1의 1이 맨 위에 옵니다.
  • Frame 4
    • EN: heappop: the top, 1 from q1, is the earliest front left. The last entry, 4, takes the top and in the same step sifts down 1 swap, past 2.
    • KO: heappop: 맨 위의 q1의 1이 남은 앞 노드 중 가장 이릅니다. 마지막 항목 4가 맨 위를 채운 뒤 같은 단계에서 자리 바꿈 1번으로 2 아래로 내려갑니다.
  • Frame 6
    • EN: 1 has a next in q1, 3, so heappush puts 3 at the end of the heap, where it stays under 2.
    • KO: q1에서 1 뒤에 3이 있으므로 heappush로 힙의 맨 끝에 넣고, 2 아래 그대로 둡니다.
  • Frame 7
    • EN: heappop: the top, 2 from q0, is the earliest front left. The last entry, 3, takes the top and stays there.
    • KO: heappop: 맨 위의 q0의 2가 남은 앞 노드 중 가장 이릅니다. 마지막 항목 3이 맨 위를 채우고 그대로 있습니다.
  • Frame 9
    • EN: 2 has a next in q0, 5, so heappush puts 5 at the end of the heap, where it stays under 3.
    • KO: q0에서 2 뒤에 5가 있으므로 heappush로 힙의 맨 끝에 넣고, 3 아래 그대로 둡니다.
  • Frame 10
    • EN: heappop: the top, 3 from q1, is the earliest front left. The last entry, 5, takes the top and in the same step sifts down 1 swap, past 4.
    • KO: heappop: 맨 위의 q1의 3이 남은 앞 노드 중 가장 이릅니다. 마지막 항목 5가 맨 위를 채운 뒤 같은 단계에서 자리 바꿈 1번으로 4 아래로 내려갑니다.
  • Frame 12
    • EN: 3 has a next in q1, 7, so heappush puts 7 at the end of the heap, where it stays under 4.
    • KO: q1에서 3 뒤에 7이 있으므로 heappush로 힙의 맨 끝에 넣고, 4 아래 그대로 둡니다.
  • Frame 13
    • EN: heappop: the top, 4 from q2, is the earliest front left. The last entry, 7, takes the top and in the same step sifts down 1 swap, past 5.
    • KO: heappop: 맨 위의 q2의 4가 남은 앞 노드 중 가장 이릅니다. 마지막 항목 7이 맨 위를 채운 뒤 같은 단계에서 자리 바꿈 1번으로 5 아래로 내려갑니다.
  • Frame 15
    • EN: 4 has a next in q2, 6, so heappush puts 6 at the end of the heap, where it stays under 5.
    • KO: q2에서 4 뒤에 6이 있으므로 heappush로 힙의 맨 끝에 넣고, 5 아래 그대로 둡니다.
  • Frame 16
    • EN: heappop: the top, 5 from q0, is the earliest front left. The last entry, 6, takes the top and stays there.
    • KO: heappop: 맨 위의 q0의 5가 남은 앞 노드 중 가장 이릅니다. 마지막 항목 6이 맨 위를 채우고 그대로 있습니다.
  • Frame 18
    • EN: 5 has a next in q0, 8, so heappush puts 8 at the end of the heap, where it stays under 6.
    • KO: q0에서 5 뒤에 8이 있으므로 heappush로 힙의 맨 끝에 넣고, 6 아래 그대로 둡니다.
  • Frame 19
    • EN: heappop: the top, 6 from q2, is the earliest front left. The last entry, 8, takes the top and in the same step sifts down 1 swap, past 7.
    • KO: heappop: 맨 위의 q2의 6이 남은 앞 노드 중 가장 이릅니다. 마지막 항목 8이 맨 위를 채운 뒤 같은 단계에서 자리 바꿈 1번으로 7 아래로 내려갑니다.
  • Frame 22
    • EN: heappop: the top, 7 from q1, is the earliest front left. The last entry, 8, takes the top and stays there.
    • KO: heappop: 맨 위의 q1의 7이 남은 앞 노드 중 가장 이릅니다. 마지막 항목 8이 맨 위를 채우고 그대로 있습니다.
  • Frame 25
    • EN: heappop: the top, 8 from q0, is the only front left, so the heap is empty after it.
    • KO: heappop: 맨 위의 q0의 8이 하나 남은 앞 노드이므로, 꺼내면 힙이 빕니다.

Closes #432
Closes #431
Closes #426
Closes #425
Closes #424
Closes #405
Closes #403
Closes #398
Closes #388

🤖 Generated with Claude Code

Kyu-seok and others added 30 commits November 20, 2026 23:30
The test file defined psql_admin five times, each answering some of
the queries and nothing for the rest. One fake now answers from the
fixture directory each block fills ($fx/databases, owned, db-state),
and a query it doesn't know, or whose fixture file is missing, ends
the run with a FAIL line naming it.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
After each fetch a clean master behind origin/master is fast-forwarded,
and the run starts again from scripts/ship-branch.sh when that changed
it; a checkout that can't be fast-forwarded and has a different script
is refused with the command that updates it. Several branches ship in
order, stopping at the first failure and naming those not reached.

CLAUDE.md drops the clause that restated the 72-character refusal.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…admin fake (#426)

Co-Authored-By: Claude <noreply@anthropic.com>
…ip several branches (#431)

Co-Authored-By: Claude <noreply@anthropic.com>
One LEFTOVER line with the count of what dev-branch-servers.sh orphans
lists, and the command. Quiet when there is nothing or Postgres is down;
a stopped Kafka or Redis is not asked.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Half of the file's 27 s was the stub backend's shutdown(): it returns at
serve_forever's next poll, 0.5 s by default, once per test. The stub now
polls every 10 ms. Another 5 s was JUDGE_RUN_SPACING=0.3 in tests that
assert nothing about spacing; only the spacing test keeps it, the rest
wait 0. What is left (about 9 s) is one Python child per Run.

Still 30 tests.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…anch databases left behind (#424)

Co-Authored-By: Claude <noreply@anthropic.com>
The pointer keeps the two rules whose breach harms other sessions and
names the cases that send an agent to docs/agents/parallel-sessions.md.
The new file adds a part on the shared compose containers. The Agent
Brief convention now says where a brief sends a doc note.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…hey take (#425)

Co-Authored-By: Claude <noreply@anthropic.com>
…moves behind a pointer (#405)

Co-Authored-By: Claude <noreply@anthropic.com>
scripts/python312.sh prints a Python 3.12 or newer (CODEBITE_PYTHON312,
python3.12, python3, each asked its version) or exits 2. judge-local.py,
replay-local.sh, replay-judge0-python.test.py and check-tracer-helpers.py
drop their own lookups; replay-local.sh no longer hands an unasked
interpreter or a /opt/homebrew guess to the replay.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Every deploy rolled the backend and recreated nginx and the worker,
whatever the merge changed. Under Docker 29's containerd image store an
image's id is the digest of an index holding the image manifest and
BuildKit's provenance attestation, and the attestation is new on every
build, fully cached or not. `provenance: false` in the compose file was
meant to stop it, but compose 5.1.0 leaves a false out of the bake
definition it hands buildx (only true or a mode gets through), so the
default attestation was still attached.

deploy-prod.sh now exports BUILDX_NO_DEFAULT_ATTESTATIONS=1 for every
build it starts, and the three dead `provenance: false` keys go.

Loop: two `docker compose build` runs of the same tree under Docker
29.3.0 and compose 5.1.0, compared by image id; and build, up, build,
up, watching for a Recreate.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…, so unchanged services stay (#403)

Co-Authored-By: Claude <noreply@anthropic.com>
The crash from #392, reproduced: open() wrote a pass's language and
theme into localStorage on the home page, where the last pass's
signed-in app was still running. When that app's GET /auth/me answered
after the write (a backend under load) and before the problem page
replaced it (a Vite under load), it wrote the user's language (ko) back.
The signed-out English pass then waited for "Description" on a Korean
page, and atTab read a tab that wasn't there: TypeError, stack trace.

open() now sets the visitor up in a script Chrome runs in the new
document before the app's, and goes straight to the problem page.
What a pass needs that doesn't show within WALKTHROUGH_WAIT_MS fails
one check naming the pass and what it waited for, with a shot of that
moment; the pass stops and the next starts. Any other error is a failed
check too, never a stack trace. The 15 s bound stays: under a Gradle
test run the slowest first load was 4.2 s.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…runs out fails a named check (#432)

Co-Authored-By: Claude <noreply@anthropic.com>
n-queens and list-bracket-pairings are drawn in full until the idea is shown, then one frame per further result; restore-wifi-ip draws a pruned branch as one frame; merge-cafe-queues draws a heap push or pop with its sift as one frame. Each folding frame's caption says what was skipped and how much.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…ong runs (#388)

Co-Authored-By: Claude <noreply@anthropic.com>
judge-local.py and check-tracer-helpers.py each stripped 'python312: ' from the finder's stderr; the line now comes bare, for a caller to put after its own name.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
dev-status counted the indented lines of orphans' human output. dev-branch-servers.sh orphans --count prints the number of databases alone; dev-status stays quiet when it prints none.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
The blank page and the images of the Diagram-beside shot were waited for with the result dropped; they now stop the section like any other wait. shown takes the passed check's line, so no check(true) follows a wait that throws; the lines printed are the same.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
restore-wifi-ip's one-call dead ends and merge-cafe-queues' pops and pushes say how much was folded (calls and lengths tried, swaps). restore-wifi-ip takes a dead end's counts from place itself, run with its frames held back, in place of a second copy of the search. The heap-move helper three tracers each carried is codebite.heap_moves; guild-chat-relay and rising-water-path regenerate unchanged. The skill names scripts/python312.sh for finding a Python 3.12 (#398).

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
For what it refuses and how it keeps itself current.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…per, named waits

Co-Authored-By: Claude <noreply@anthropic.com>
@Kyu-seok
Kyu-seok marked this pull request as ready for review November 20, 2026 15:57
@Kyu-seok
Kyu-seok merged commit e78da5b into master Nov 22, 2026
10 checks passed
@Kyu-seok
Kyu-seok deleted the integration/tickets-2026-11-20 branch November 22, 2026 05:13
Kyu-seok added a commit that referenced this pull request Nov 22, 2026
* SessionStart hook: fast-forward a stale main checkout

A session reads CLAUDE.md, skills and scripts from the main checkout, which
nothing kept current: it sat seven merges behind through the #433 build.
The hook fast-forwards a clean master and says what it moved, or says how
far behind the checkout is and the command that catches it up.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

* Tracer helper check: compare nested functions too

A third copy of a nested heap helper reached review in #433: the check
compared only top-level functions. The twelve nested copies already there
are listed in nested-copies.known, which passes exactly those and fails on
a new copy or a stale line; #435 moves them into codebite.py.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5.2 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment