Python 3.11.0下无法通过Poetry安装scikit-learn的问题

环境配置 #

poetry: 1.2.2
pyenv: 2.3.5-3-g0726e02e
python: 3.11.0
目标安装版本scikit-learn: 1.1.2

问题现象 #

执行poetry add scikit-learn时触发元数据生成错误，核心报错如下：

Preparing metadata (pyproject.toml): still running...
Preparing metadata (pyproject.toml): finished with status 'error'
error: subprocess-exited-with-error

× Preparing metadata (pyproject.toml) did not run successfully.
│ exit code: 1
╰─> [273 lines of output]
    Partial import of sklearn during the build process.
    setup.py:128: DeprecationWarning:
    
      `numpy.distutils` is deprecated since NumPy 1.23.0, as a result
      of the deprecation of `distutils` itself. It will be removed for
      Python >= 3.12. For older Python versions it will remain present.
      It is recommended to use `setuptools < 60.0` for those Python versions.
      ....
      ....
      ....
    self.finalize_unix()
        File "/tmp/pip-build-env-go8f9tm1/overlay/lib/python3.11/site-packages/setuptools/_distutils/command/install.py", line 498, in finalize_unix
          self.select_scheme("posix_prefix")
        File "/tmp/pip-build-env-go8f9tm1/overlay/lib/python3.11/site-packages/setuptools/_distutils/command/install.py", line 528, in select_scheme
          return self._select_scheme(resolved)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        File "/tmp/pip-build-env-go8f9tm1/overlay/lib/python3.11/site-packages/setuptools/_distutils/command/install.py", line 537, in _select_scheme
          setattr(self, attrname, scheme[key])
                                  ~~~~~~^^^^^
    KeyError: 'headers'
    [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

已尝试操作 #

• 执行poetry run pip install setuptools==59.8.0降级setuptools，但再次执行poetry add scikit-learn时，Poetry会自动将setuptools升级至65.0.x版本，导致安装失败。
• 环境中单独安装numpy、pandas、scipy等库无问题，无需降级setuptools。

解决方法 #

方法1：锁定setuptools版本到pyproject.toml #

1. 在项目的pyproject.toml中添加setuptools的版本锁定：

[tool.poetry.dependencies]
# 保留你已有的其他依赖，新增以下行
setuptools = "==59.8.0"

2. 执行poetry lock --no-update生成锁定文件，避免Poetry自动升级setuptools。
3. 最后执行poetry add scikit-learn==1.1.2完成安装。

方法2：指定构建环境的setuptools版本 #

在pyproject.toml中添加构建系统配置，强制构建时使用指定版本的setuptools：

[build-system]
requires = ["setuptools==59.8.0", "wheel"]
build-backend = "setuptools.build_meta"

配置完成后直接执行poetry add scikit-learn==1.1.2即可。

方法3：升级Poetry版本 #

你当前使用的Poetry 1.2.2版本较旧，可能存在依赖处理逻辑缺陷。执行以下命令升级到最新稳定版：

poetry self update

升级后再尝试poetry add scikit-learn==1.1.2，新版本Poetry可能自动兼容这类依赖冲突。

方法4：直接安装预编译轮子包 #

如果上述方法都无效，绕开元数据构建流程，直接用pip安装预编译的scikit-learn轮子包：

poetry run pip install scikit-learn==1.1.2 --only-binary :all:

内容的提问来源于stack exchange，提问作者winter