detect_changes: report the code a changed documentation section references
メンテナーはふだん 2 日以内に返信
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
- 35/100
調査の方向性
Start with detect_is_seedable_label and detect_walk_parse in src/mcp/mcp.c, then read doclink_md.c to understand how references are attributed to sections. Reproduce with case 5 from the script attached to #2585 and inspect how prose edits affect section ownership and MENTIONS traversal. Done means the design questions in the issue are resolved and changed documentation can report its referenced code separately from blast-radius results.
索引モデルが issue の本文から書いたものです。
説明
Concerns detect_changes together with the doc-links feature from #2551. I
found no existing issue or PR.
A documentation-only edit produces no seeds, so detect_changes cannot list
the code a changed section references, although the graph already holds the
MENTIONS edges that do:
MENTIONS from the section: [['Posting', 'post_entry']]
doc-only edit, edge_types ["CALLS","MENTIONS"]: changed_files=['docs/notes.md'] seed_symbols=0
detect_is_seedable_label (src/mcp/mcp.c) excludes Section, along with
File, Folder, Module, Project and Package. That is the right default
for the blast-radius walk. The code-to-doc direction already works: a
code-only edit to post_entry reaches the Posting section at hop 2 with the
same call. Only the doc-to-code direction is missing.
What this would report. The code a changed section names: code that may
need review because its documentation changed. That is not behavioral impact.
Changing a document changes nothing the code does, so this output should be
labeled apart from the blast radius, not merged into it.
Simply making Section seedable is not enough. There are three design
questions I'd like settled before sending a PR:
- Which section owns a changed line? A Section node's stored range covers
its heading line only, not the prose below it (push_simple_class_def_doc
records the heading node's lines). Ordinary hunk intersection would miss
edits to the prose. Since nothing would overlap, it would also trip the
whole-file fallback and seed every section of the document. One option: a
changed line belongs to the nearest heading at or above it, the same rule
doclink_md.cuses to attribute references. Another: store each Section's
body range on the node. - Traversal direction. A request with
edge_typesmust beinbound
(detect_walk_parserejects anything else: "edge_types, fixpoint and via
walk the blast radius only"). Reaching code from a Section means following
MENTIONSoutbound. Should this be an explicit mode ofdetect_changes
(for exampledoc_references: true)? Or a separate doc-to-code step, one
outboundMENTIONShop from the changed sections, whose results are
reported, and possibly fed to the inbound walk as further seeds? - File-level references. Should a change above a document's first heading
use the document's File node? doc_links attributes references written there
to the File.
Environment: built from main at ec4fbd3a (2026-10-10), Linux x86-64, CLI
mode, fresh cache.
Reproduction
CBM="codebase-memory-mcp cli" ./repro.sh 5, using the script attached to
#2585. Case 5 indexes docs/notes.md, which has a ## Posting section that
references src/ledger.py::post_entry. It edits only the prose and re-indexes,
then calls detect_changes with edge_types: ["CALLS","MENTIONS"].
- 主要言語
- C
- スター
- 45.8k
- フォーク
- 3.8k
- 平均マージ
- 3日 6時間
- マージ済み PR（30日）
- 201
環境構築
このプロジェクトには開発コンテナ、Dockerfile、コントリビューションガイドがありません。まず README を読み、一般的な手順ははじめてのコントリビューションガイドを参照してください。
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。
DeusData/codebase-memory-mcp のほかの issue
-
Generated Pi extension (cbmem.ts) breaks Gemini/Antigravity: Type.Union of Type.Literal serializes to anyOf/const対応中かも @divakantpandey が 1 日前に担当しました。 オープンbug editor/integration
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
DeusData/codebase-memory-mcp#2787 · コメント 1 件 ·
メンテナーはふだん 2 日以内に返信
-
editor/integration parsing/quality
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
DeusData/codebase-memory-mcp#2286 ·
メンテナーはふだん 2 日以内に返信
-
install writes an unsupported `args` key into the Claude Code SubagentStart hook, so it never runsオープンbug editor/integration priority/high stability/performance ux/behavior
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
DeusData/codebase-memory-mcp#2239 · コメント 1 件 ·
メンテナーはふだん 2 日以内に返信
-
bug priority/high ux/behavior
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
DeusData/codebase-memory-mcp#2232 · コメント 2 件 ·
メンテナーはふだん 2 日以内に返信
-
documentation editor/integration parsing/quality priority/normal
難易度 2/5 1〜3時間 初心者へのやさしさ 85/100
DeusData/codebase-memory-mcp#2169 ·
メンテナーはふだん 2 日以内に返信
DeusData/codebase-memory-mcp の issue をすべて見る
似ている issue
-
documentation
難易度 1/5 1〜3時間 初心者へのやさしさ 91/100
メンテナーはふだん 1 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 83/100
メンテナーはふだん 1 日以内に返信
-
enhancement ui
難易度 1/5 1時間未満 初心者へのやさしさ 72/100
ChupathingyCE/chupathingyce#485 ·
メンテナーはふだん 1 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 65/100
メンテナーはふだん 1 日以内に返信
-
難易度 1/5 1時間未満 初心者へのやさしさ 82/100
eclipse-paho/paho.mqtt.c#1727 ·
メンテナーはふだん 6 日以内に返信