project-chip/connectedhomeip

Parallelize code generation for jinja files and code formatting

Open

#38,775 建立於 2025年5月6日

在 GitHub 查看
 (1 留言) (0 反應) (0 負責人)C++ (6,410 star) (1,659 fork)batch import
good first issuestale

描述

We want to have fast iteration speed on code generation. https://github.com/project-chip/connectedhomeip/blob/master/scripts/py_matter_idl/matter/idl/generators/cpp/sdk/sdk_generator.py#L120 currently generates a LOT of files and could benefit from paralelization

Update code logic to do parallel generation and test out if this improves runtime (i.e. if it is worth it - this issue was created on the assumption that it is, however we should verify that).

We also do file formatting and those seem to call one single exec to format many files at once. We should investigate if splitting files per cpu count improves formatting (i.e. is clang-format and kotlin-format single threaded or multi-threaded?)

貢獻者指南

技術棧
pythoncppkotlin
領域
toolingbuild system
議題類型
performance
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
3
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
3-5 days
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
fresh
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
mostly clear
前置要求
PythonConcurrency basicsBuild system
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
45
研究方向
Investigate the file scripts/py matter idl/matter/idl/generators/cpp/sdk/sdk generator.py at line 120 to understand the current sequential generation. Implement parallel generation using Python's concurrent.futures module. Measure runtime before and after to confirm improvement. Additionally, examine clang format and kotlin format to determine if they are single threaded; if so, split the file list by CPU count and format in parallel. Report the performance results in the issue.