Work plan — QuantEcon/actions: wave 7 (#300, #27) and v0.18.0, the first gated release

Aperta
  • Qualcosa non va?
#305 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
1/100
Tipo di issue
Funzionalità
Chiarezza
Da chiarire
Stato di attività
Attiva
Stack tecnologico
github-actions, shell
Ambito
devops

Direzione di ricerca

This issue is a live work plan covering release #304, the release gate, wave 7 checks for issues #300 and #27, and several maintainer decisions. Start by reading the linked issues and PRs, then check the current repository and workflow state against the plan’s “Where we stand” section. The plan spans parallel work, release coordination, and approval dependent decisions, so it does not describe a bounded first contribution.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

Continuing from #246, closed on 2026-11-10 with its ledger. This plan covers wave 7 and v0.18.0, the first release under the release gate. It is worked by parallel Claude Code sessions, each owning one area and one pull request at a time, with an organiser session coordinating them. This body is the single source of truth for the work. Revise it in place against live state before working from it.

Where we stand (verified 2026-11-10 21:32 UTC)

What State Measured
main 2ed1629 (#293) git ls-remote
Latest release v0.17.0, tag on fae622c git ls-remote
v0 fae622c (v0.17.0), moved 2026-11-10 16:18 UTC with the maintainer's approval, after the lecture matrix (run 42764767724, 8/8) and the release gate (canary run 42776388278) were both green git ls-remote
v0.18.0 Release PR #304 is open: preview-workers (#298), the GHCR pruner (#290), the release-gate docs (#301), and Dependabot's #294, #295 and #293 gh pr list
The release gate Live: gate.yml in QuantEcon/test-actions-lecture-intro, with its fixture in gate/. CONTRIBUTING's Staged Releases uses it (#301). Proving runs on the canary's main: v0.16.0 green, both self-tests red where they should be, v0.17.0 green canary runs
Container images Rebuilt and promoted twice on 2026-11-10, for #295 and #293 (theme 0.22.2). Lecture-matrix run 42880336347 tests them: 7 of 8 legs green, the last running gh run view
Open PRs #304 (v0.18.0), #279 (the maintainer's CHANGELOG footer fix; theirs to merge) gh pr list
Sessions The organiser, mini:actions. Wave 7's mini:actions-build and mini:actions-restore launched 2026-11-10 19:30 UTC, and are doing their issue checks tmux

Front of the plan:

  1. v0.18.0: merge #304, tag, Release. Then the gate on v0.18.0, and the hand-staged legs for what it doesn't reach (preview-workers, the previews' pull-request half, publish-gh-pages' release assets), with cache.yml once side by side. v0 moves only when all of that is green, with approval.
  2. Wave 7's checks, as they arrive: proposal-checks/300.md and proposal-checks/27.md. Each question goes to the maintainer one at a time, and the agreed check is posted on its issue.
  3. Lecture-matrix run 42880336347. If it goes red, find which image change is to blame before anything else merges into containers/**.
  4. The lecture sites' first runs on the new v0 and the new images, from 2026-11-11.

Wave 7

Item Session State
#300: a cell that raises passed -W --keep-going in a cache-mode jupyter build; only the error-output check caught it. A caller with error-outputs: warn or ignore would publish it mini:actions-build Issue check
#27: a restore-gh-pages action and templates/restore.yml, the inverse of publish-gh-pages (home decided 2026-09-25) mini:actions-restore Issue check

Neither PR goes in v0.18.0.

Decisions waiting on the maintainer

Issue Question What it unblocks
#97 HTML strictness across the org (-n, and extra-args through build-jupyter-cache) The full chain for lecture-python-intro and lecture-python-programming (PLAN's rollout rows 3 and 4)
#194 Should link checking and Colab checks move into shared actions? #184, #185
#195 Build sync-notebooks (#96), or retire the .notebooks repositories first? #96

Releases

  • v0.17.0: shipped 2026-11-10. v0 is on it.
  • v0.18.0: #304, the first release staged through the gate.
  • After v0.18.0: wave 7's PRs, and #302 (scheduling the canary's ci.yml and publish.yml, PLAN P0 item 0c), which waits until the gate has replaced the hand-staged legs.

Follow-ups carried from #246

  • The image-size report is not exactly reproducible. The lean image's on-disk figure varied by 111,355 bytes with identical layers.
  • publish-gh-pages and a / in the tag. Under mode: all it fails only at the release step, after the deploy. Left for the maintainer.
  • The two deploy-cloudflare callers are pinned to v0.12.0 and pass neither visibility nor require-access. Each needs visibility: private when it moves to v0.16.0 or later.
  • The lecture family's move to @v0 is unblocked on this side (the note on QuantEcon/workspace-lectures#33). When to repin is the lecture family's call.
  • #299: preview-workers' pilot checks wait for the first lecture repository that adopts it.
  • Low priority: #291, #282, #302, and #311 (build-jupyter-cache's jupyter repro command exits 0, found during #300's work).

Explicitly not doing

  • #147's template items (the concurrency fallback, paths-ignore).
  • #150 and #151 while a lecture-matrix run is going, since both rebuild the images.
  • #148, #149 and #299 here: they are work in the lecture repositories.
  • #18's caching, parked until a repository passes environment-update.
  • Committing the 2026-11-04 audit bundle to .qe/dev/log/. That is the maintainer's call.

Resuming

The organiser session is mini:actions, and the workers are mini:actions-<name>. The briefs live in the maintainer's checkout under .claude/briefs/ and are not committed: HANDOVER.md is the organiser's start-up file; common.md holds the worker rules and <name>.md each worker's brief; proposal-checks/ holds each issue's check; release-templates/ holds the last release's PR body, notes and canary files. A fresh session can work from this issue, the merged PRs and the decision records alone.

Claude Code (on behalf of mmcky), organiser session

Lingua principale
Shell
Stelle
0
Fork
1
Merge medio
18h 51m
PR unite (30g)
58

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di QuantEcon/actions

Tutte le issue di QuantEcon/actions

Issue simili

Altre issue su Shell/Bash

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.