bazel-contrib / bazel-contrib/rules_python

Allow using a locally installed Python runtime as a full toolchain

オープン
#2,070 コメント 7 件 リアクション 0 件 担当者 0 名 GitHub で見る
主要言語
Starlark
スター
688
フォーク
721
平均マージ
15時間 7分
マージ済み PR(30日)
76

説明

This feature request has come up a few time from different places.

The basic request is to have an option other than the hermetic runtimes, typically a Python runtime that is locally installed. Some of the reasons this is appealing:

* Avoids having to download a remote runtime.
* If you deeply care about supply chain issues, the provenance of the runtimes isn't very compelling.
* The standalone-python-build runtimes have some known limitations
* The hermetic runtimes are expensive to setup: they're inbuild runtimes with a (mostly) full python installation, all of which has to be copied into runfiles
* I think some people like to manage their Python version outside of bazel, e.g. using `pyenv` or venv activation.

Making a local runtime have the same capabilities as a hermetic runtime is _sort of_ possible today, but requires quite a bit of setup, and I think maybe loading a few private bzl files.

PR #2000 lays the ground work for most of it. What's now missing is a public API for it and figuring out what knobs (if any) it needs.

For at least early testing, we're thinking to have a flag control what type of runtime is selected.

But, during our maintainers meeting, two big questions came up:
1. What does the public API for using it look like?
2. What are the use cases? These drive what the public API needs to look like

So, if you have use cases, please post them here. It'll help us figure out what capabilities are needed.

TODO
* [x] Create repo rules to map the local python install into bazel
* [x] Allow setting arbitrary target_compatible_with/target_settings
* [ ] Allow creating an in-build instead of platform runtime

コントリビューションガイド

コントリビューションガイドを開く

調査の方向性

まず PR #2000 と、ローカルの Python インストールを Bazel にマッピングするために完了した TODO 項目を確認します。文書化されたユースケースに基づいて公開 API と必要な設定項目を定義し、そのうえで、プラットフォームランタイムではなく in-build のものを作成する方法を決定します。

索引モデルが issue の本文から書いたものです。

評価

技術スタック
python
領域
build-system
issue の種類
機能追加
難易度
5/5
見積もり時間
1週間以上
活発さ
停滞
明瞭さ
説明が足りない
初心者へのやさしさ
25/100

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。