[Phase 16] S13 — Packaging metadata, dependency single-source, container distribution
まだ誰も着手していません。
評価
- 難易度
- 5/5
- 見積もり時間
- 1週間以上
- 初心者へのやさしさ
- 12/100
- issue の種類
- リファクタリング
- 明瞭さ
- おおむね明確
- 活発さ
- 活発
- 技術スタック
- docker, docker-compose, python
調査の方向性
Blocked by #888, so start there; this step is reviewed against that standard. Begin with the roadmap plan at docs/roadmap/phase-16-trust-surface-hardening.md lines 682-721, then the cache mount in docker-compose.yaml (hf_cache at /root/.cache/huggingface versus USER forgelm). Done means the seven exit criteria pass, including the workflow_dispatch smoke job and twine check.
索引モデルが issue の本文から書いたものです。
説明
Roadmap · Phase 16, step S13 · size L · 7 unit(s) · planned in
docs/roadmap/phase-16-trust-surface-hardening.md:682-721
Summary
The distribution surface — what a user installs and what they run it in — with one root cause: hand-maintained duplicates of a single truth. docker-compose.yaml mounts hf_cache:/root/.cache/huggingface while the image drops to USER forgelm, so the cache never persists (an artifact defect, not only a documentation one). The manual advertises three image variants (latest, slim, airgap) that nothing builds, and every documented docker run … forgelm --config … resolves to forgelm forgelm … against the shipped ENTRYPOINT.
Units
| Unit | Tracked by (October 2026 review) |
|---|---|
C31-DOCKER-RUNBOOK (= DOCS-05 + OPS-13) |
#157 · #159 · #201 · #210 · #217 · #218 · #298 · #409 · #422 · #451 · #602 · #615 · #619 · #638 · #643 · #702 · #789 |
F-W20260729-OPS-11, F-W20260729-OPS-12, F-W20260729-OPS-14, F-W20260729-OPS-17, F-W20260729-OPS-18 |
#114 · #465 · #523 · #558 · #566 · #607 · #668 · #686 · #702 · #776 |
C31-DEPENDENCY-SOURCE (= OPS-07 + OPS-08) |
#537 · #686 · #702 |
Blocked by
#888 — the standard this step is reviewed against; write it first.
Decisions this step executes
- C-15 — container distribution: local
docker buildis the contract; a published image is tracked as a condition (deferred issue). - C-16 — fix the
docker-compose.yamlcache path in the same change, with a### Fixedentry naming the orphaned-volume effect.
Plan
- Delete the three phantom variants and re-derive README and both manuals from the artifacts on disk.
- Fix the compose cache path in the same change.
Exit criteria
- README and both manuals describe exactly one distribution path.
- A guard fails when a first-party registry reference has no backing publish workflow, when a compose file name in prose is absent from disk, or when a documented service name is absent from
services:. - The documented first container command succeeds in a
workflow_dispatchsmoke job. - Two disposable
docker compose run --rminvocations share one cached snapshot without root. - The documented wizard command leaves a validated YAML on the host after
--rmthat passespython3 -m forgelm --config <file> --dry-run. - A meta-test compares normalised direct dependencies between the nightly minimum-deps list and the package metadata.
- A clean isolated build emits no packaging-metadata deprecation warning and
twine check dist/*passes.
How to deliver
- One root-cause family, committed on its own; then an Opus review round and a Sonnet review round, each with verified findings fixed and committed before the next step begins.
- A reproducer that is red before the fix and green after; the full gauntlet from
CLAUDE.mdpasses under./.venv/bin/python. - Every English documentation edit ships with its Turkish mirror in the same commit; a new or renamed audit event gets its EN and TR catalog rows in the same PR.
- A step that grows a deferred module carries its
budget_historyjustification in the same diff; a widened guard rolls out non-strict first, then flips. - Commit bodies cite
Absorbs F-W20260729-…; the CHANGELOG[Unreleased]section gets an entry only for a change a user or library consumer can observe (decision C-20).
Planned in docs/roadmap/phase-16-trust-surface-hardening.md (read at f94595f). Unit IDs (F-W20260729-…) come from the 2026-07-29/30 full-project review; the issue numbers next to them are the October 2026 review's records of the same defects.
- 主要言語
- Python
- スター
- 9
- フォーク
- 1
- 平均マージ
- 8時間 10分
- マージ済み PR(30日)
- 5
環境構築
- Dockerfile または Docker Compose ファイルあり
- プルリクエストのテンプレートあり
- コントリビューションガイドを読む
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
HodeTech/ForgeLM のほかの issue
-
area: dev-tooling bug severity: low source: roadmap wave: 4
難易度 2/5 1〜3時間 初心者へのやさしさ 70/100
-
area: site bug good first issue severity: low source: review-2026-09 wave: 4
難易度 2/5 1〜3時間 初心者へのやさしさ 66/100
-
documentation good first issue severity: medium source: review-2026-09 wave: 3
難易度 2/5 1〜3時間 初心者へのやさしさ 75/100
-
documentation good first issue severity: low source: review-2026-09 wave: 4
難易度 1/5 1〜3時間 初心者へのやさしさ 80/100
-
documentation severity: medium source: review-2026-09 wave: 3
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
HodeTech/ForgeLM の issue をすべて見る
似ている issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 76/100
spec-kitty/spec-kitty#6112 ·
メンテナーはふだん 1 日以内に返信
-
[Bug]: OpenRouter streaming chunks report the requested router name instead of the model OpenRouter served対応中かも @Mrun25 が今日担当しました。 オープンbug llm translation
難易度 2/5 1〜3時間 初心者へのやさしさ 75/100
BerriAI/litellm#46095 · コメント 1 件 ·
メンテナーはふだん 1 日以内に返信
-
bug
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
uberdeveloper/fastbt#29 ·
-
bug
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
andrew-blake/melcloudhome#368 ·
メンテナーはふだん 1 日以内に返信
-
lane:discover
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
SPARDiffData/differential-data-collection#33 ·
メンテナーはふだん 1 日以内に返信