Hello. This time I’ll introduce python-build-standalone, which enables deployment of portable Python environments supporting Linux, macOS, Windows/x86, and ARM.

GitHub - astral-sh/python-build-standalone: Produce redistributable builds of Python

Produce redistributable builds of Python. Contribute to astral-sh/python-build-standalone development by creating an account on GitHub.

github.com

Overview

The python-build-standalone repository published on GitHub distributes pre-built Python binaries for various OS and CPU architectures. The distributed Python includes pip, so you can start using the Python environment immediately after download. Additionally, the Python provided here is used by recently notable package managers like Rye and mise, helping to build flexible Python environments.

Due to licensing, the distributed Python may behave differently from self-built versions, but there are no issues in most operations. Details are written in the following article:

Ryeに見る自己完結型Pythonとライセンスの話 - techtekt

目次 目次 はじめに 検証環境 pyenv + Poetryのインストールとプロジェクト作成 pyenvのインストール Poetryのインストール プロジェクト作成 Ryeのインストールとプロジェクト作成 インストール プロジェクト作成 pyenv環境とRye環境のPythonの挙動の違い readlineモジュール なぜRyeのPythonはlibeditを使用するのか pyenvよるPythonのインストール RyeによるPythonのインストール ライセンスの問題 ユーザへの影響 まとめ はじめに Pythonの開発において、pyenvによるバージョンの切り替えと、Poetryによるプロ…

techtekt.persol-career.co.jp

Provided Python

python-build-standalone provides pre-built Python binaries for various OS and CPU architectures. They can be obtained from the following GitHub Release page:

Releases · astral-sh/python-build-standalone

Produce redistributable builds of Python. Contribute to astral-sh/python-build-standalone development by creating an account on GitHub.

github.com

The naming convention for available binaries follows this format:

cpython-(Python version)+(release timestamp)-(architecture)-(build configuration)-(included content)

Each item is as follows:

Release Timestamp

Enter the date when the Release was published, set as a GitHub Tag. The latest Release date can be confirmed from the following GitHub API:

raw.githubusercontent.com

raw.githubusercontent.com

Architecture

SIMD-enabled x86_64_v2-*, x86_64_v3-*, x86_64_v4-* will crash if run on unsupported CPUs. According to documentation, *-unknown-linux-gnu is recommended for Linux, *-windows-msvc-shared for Windows. For macOS, choose according to CPU.

Other architectures like armv7 not mentioned here are also included in actual Releases, so please check.

Build Configuration

• pgo+lto: Profile guided optimization and Link-time optimization applied during build. Fastest distribution
• pgo: Only Profile guided optimization applied
• lto: Only Link-time optimization applied
• noopt: Build with only normal optimization
• debug: Debug build without optimization

Included Content

• install_only: Only files needed for post-build installation
• install_only_stripped: Lightweight version of install_only with debug symbols removed
• full: All files and artifacts used in build included. Distributed in .tar.zst format

Let’s look at capacity differences between the above:

Usage with Docker

Let’s try introducing Python obtained from python-build-standalone in a Dockerfile.

This time, I’ll install cpython-3.11.9+20240726-x86_64-unknown-linux-gnu-pgo+lto-full.tar.zst on a debian:latest image.

FROM --platform=linux/amd64 debian:latest

ARG TAG="20240726"
ARG PYTHON_VERSION="3.11.9"

RUN rm -f /etc/apt/apt.conf.d/docker-clean \
    && echo 'Binary::apt::APT::Keep-Downloaded-Packages "true";' > /etc/apt/apt.conf.d/keep-cache
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
    --mount=type=cache,target=/var/lib/apt,sharing=locked \
    apt-get update && apt-get install -y --no-install-recommends \
    curl \
    ca-certificates \
    zstd

RUN curl -sSL "https://github.com/indygreg/python-build-standalone/releases/download/${TAG}/cpython-${PYTHON_VERSION}+${TAG}-x86_64-unknown-linux-gnu-pgo+lto-full.tar.zst" -o python.tar.zst \
    && tar -axvf python.tar.zst \
    && mv python/install /usr/local/python \
    && rm -rf python.tar.zst python
ENV PATH="/usr/local/python/bin:$PATH"
ENV LD_LIBRARY_PATH="/usr/local/python/lib:$LD_LIBRARY_PATH"

docker build -f Dockerfile ./ -t debian-python

docker run --rm -it debian-python python
Python 3.11.9 (main, Jul 25 2024, 22:42:09) [Clang 18.1.8 ] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>>

The Python interpreter started successfully.

To introduce arbitrary Python versions in Docker containers, options were limited to registering apt repository ppa:deadsnakes/ppa, building with pyenv, poetry, etc., or building yourself. Using python-build-standalone enables easy environment setup. Particularly, NVIDIA Docker images like nvidia/cuda don’t include Python by default, so this method allows easy Python environment addition.

Summary

This time I introduced the method of installing pre-built Python supporting various OS and architectures distributed by python-build-standalone into Docker. This method enables easy Python environment construction without new builds.