JuliaPy / JuliaPy/PythonCall.jl
Turn off conda initialization during incremental compilation?
- 主要言語
- Julia
- スター
- 1.1k
- フォーク
- 86
- 平均マージ
- 1日 22時間
- マージ済み PR(30日)
- 3
説明
If you have a `using PythonCall` inside your package "Foo", then just _precompiling_ Foo will trigger creation of the PythonCall's Conda environment, downloading of packages, etc...
IMO this is undesired. I know there's environment variables to prevent this, but while I remember to set them when actually running a program, its very often I forget during precompilation, especially since precompilation automatically gets triggered at various times (when instantiating, upgrading, etc...)
Wondering if it makes sense to turn that off during precompilation entirely with a check to `ccall(:jl_generating_output, Cint, ())`? It would mean Python wouldn't be available to any dependent packages during precompilation either, but perhaps thats a better behavior? I don't have the strongest feelings one way or the either, probably leaning slightly that it'd be nicer to make this change, but just wanted to point out.
コントリビューションガイド
このリポジトリのコントリビューションガイドは索引されていません
調査の方向性
Start by reviewing how PythonCall handles environment initialization during Julia package precompilation, including the existing environment-variable controls and the proposed ccall(:jl_generating_output, Cint, ()) check. Determine the intended behavior for dependent packages that need Python during precompilation, then verify the chosen behavior with precompilation scenarios.
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
- julia, 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
- おおむね明確
- 初心者へのやさしさ
- 25/100