After the classic Converse deletion (#4061): dangling references and decisions on the Charm++ side
メンテナーはふだん 1 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
- 25/100
- 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
- cmake, cpp, linux
調査の方向性
Start by checking the listed callers and headers: src/libs/ck-libs/state_space_searchengine/searchEngine.C, src/ck-core/ck.C, src/ck-core/init.C, and src/conv-core/debug-conv.h. Compare each reference with the deleted classic implementation and the linked reconverse issues; the issue leaves several choices open, so first establish which decisions are in scope. Done means resolving the dangling references and recording decisions for the unsupported paths, including the charmc link flags.
索引モデルが issue の本文から書いたものです。
説明
Companion to #4061. The inventory of the 699 deleted files, grouped into 45 capabilities with a disposition for each, is at https://claude.ai/artifact/XuUvYjLttB9Ntz7XuMUjkd (section "Capabilities removed, and where each one stands"). This issue holds the Charm++-side items that still reference a deleted capability or need a decision; reconverse #296 holds the reconverse-side ones.
Dangling references to decide
- CmiPoolAlloc: src/conv-core/cmipool.h is kept but cmipool.C is deleted. src/libs/ck-libs/state_space_searchengine/searchEngine.C still calls CmiPoolAllocInit(30). Either port the pool into reconverse or switch that library to CmiAlloc and drop the header.
- Message compression: reconverse never compresses, and the classic compress*.C / lz4 / zlib files in arch/util are gone. The kept Charm++ tree has no caller of Converse-level message compression: the only compression references are
CMK_USE_ZLIBblocks in ck-ldb (LBManager.C, readmodel.C: gzip for load-balancing dump files, using zlib directly) and in AMPI (disabled with&& 0). Nothing to fix here unless compression is wanted back; decision in reconverse #296. - CmiStartQD in the record-replay path: src/ck-core/ck.C calls CmiStartQD for CkMessageReplayQuiescence; reconverse has no CmiStartQD. Check whether that path is compiled on reconverse, then either change the caller to CkStartQD or ask reconverse for the function.
- CharmDebug hooks: src/conv-core/debug-conv.h and debug-conv++.C are kept; src/ck-core/init.C declares and calls CpdFreeze; the classic debug-conv.C (CpdInit, CpdFreeze, CpdNotify) is deleted and reconverse has none of it. CharmDebug has not been tried on reconverse. Decide whether CharmDebug is supported on the 9.0 line.
- Debugging allocators (memory-paranoid, memory-leak, memory-record, memory-verbose, memory-charmdebug): gone with no reconverse equivalent. Sanitizers cover paranoid and leak; CharmDebug's memory views depended on memory-charmdebug. Goes with the CharmDebug decision.
- Out-of-core (CMK_OUT_OF_CORE): decide later (2026-10-20). Nothing was deleted: the repository only ever held conv-ooc.h (checked in commented-out in 2003), ten guarded lines in cklocation.C, ckarray.C and ckmigratable.h, and the
oocoption in buildcmake; no CooRegisterManager/CooRegisterObject implementation exists in the history, so the option fails at link in classic too. Options: drop the header, guards and build option, or keep them as the shape of a future design. - -rdynamic on the charmc link line for reconverse targets: reconverse #294 adds stack traces on CmiAbort and fatal signals; on Linux the application's own symbols appear in the trace only if the executable is linked with -rdynamic. reconverse's CMake interface adds it for CMake consumers; charmc-linked programs need it in the reconverse-linux-* conv-mach.sh.
Unknown
- Windows: the win and *-win-x86_64 arch directories are deleted; reconverse and LCI have never been built on Windows. cc-msvc in src/arch/common is kept.
Slated elsewhere (for the record)
Isomalloc and migratable threads, and AMPI global privatization: AMPI stages 1-2. Seed balancers and their LBTopology neighborhoods: reconverse #185 re-port. Stack traces: reconverse #294. Converse futures: reconverse #295. MPI interop: reconverse #296.
🤖 Generated with Claude Code
- 主要言語
- C++
- スター
- 241
- フォーク
- 62
- 平均マージ
- 1日 1時間
- マージ済み PR(30日)
- 125
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。
charmplusplus/charm のほかの issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
charmplusplus/charm#4210 ·
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 76/100
charmplusplus/charm#4160 ·
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 70/100
charmplusplus/charm#3619 · コメント 4 件 ·
メンテナーはふだん 1 日以内に返信
-
CkFreeSysMsg pools non-system messages; a later CkAllocSysMsg can return a CkReductionMsg対応中かも @lvkale が今日担当しました。 オープン
難易度 3/5 1〜2日 初心者へのやさしさ 35/100
charmplusplus/charm#4224 ·
メンテナーはふだん 1 日以内に返信
-
難易度 4/5 3〜5日 初心者へのやさしさ 48/100
charmplusplus/charm#4223 ·
メンテナーはふだん 1 日以内に返信
charmplusplus/charm 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 69/100
-
area:qmc bug
難易度 2/5 1〜3時間 初心者へのやさしさ 69/100
メンテナーはふだん 3 日以内に返信
-
bug good first issue
難易度 2/5 1〜3時間 初心者へのやさしさ 74/100
ASGC-Game-Jam/asgc-gamejam-2026#563 ·
メンテナーはふだん 1 日以内に返信
-
area/ysql kind/bug priority/medium status/awaiting-triage
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
yugabyte/yugabyte-db#37523 ·
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 66/100