Ubuntu 24.04下PyPy安装Scipy失败问题求助

问题背景 #

使用的环境信息：

Python 3.8.13 (7.3.9+dfsg-1ubuntu0.1, Nov 15 2022, 06:22:50)
[PyPy 7.3.9 with GCC 11.3.0]

操作系统为Ubuntu 24.04，已执行以下命令完成基础安装：

• sudo apt install pypy3
• sudo apt install gfortran

执行pypy -m pip install scipy时失败，核心错误为多次出现Segmentation fault (core dumped)，具体失败环节如下：

[263/1619] Generating scipy/special/_data_gsl with a custom command
      FAILED: scipy/special/gsl.npz
      /usr2/repositories/DSP/dspos/RoomResponseRef/source/python/some_env/bin/pypy3 ../../scipy/special/utils/makenpz.py --use-timestamp gsl -o scipy/special
      Segmentation fault (core dumped)
      [264/1619] Compiling C++ object scipy/special/_ufuncs_cxx.pypy38-pp73-x86_64-linux-gnu.so.p/ellint_carlson_wrap.cxx.o
      [265/1619] Generating scipy/special/_data_boost with a custom command
      FAILED: scipy/special/boost.npz
      /usr2/repositories/DSP/dspos/RoomResponseRef/source/python/some_env/bin/pypy3 ../../scipy/special/utils/makenpz.py --use-timestamp boost -o scipy/special
      Segmentation fault (core dumped)
      [266/1619] Generating scipy/linalg/flinalg_module with a custom command
      [267/1619] Generating scipy/special/_data_local with a custom command
      FAILED: scipy/special/local.npz
      /usr2/repositories/DSP/dspos/RoomResponseRef/source/python/some_env/bin/pypy3 ../../scipy/special/utils/makenpz.py --use-timestamp local -o scipy/special
      Segmentation fault (core dumped)
      [268/1619] Generating 'scipy/special/_comb.pypy38-pp73-x86_64-linux-gnu.so.p/_comb.c'
      [269/1619] Generating 'scipy/_lib/_ccallback_c.pypy38-pp73-x86_64-linux-gnu.so.p/_ccallback_c.c'
      [270/1619] Generating 'scipy/_lib/messagestream.pypy38-pp73-x86_64-linux-gnu.so.p/messagestream.c'
      [271/1619] Generating scipy/linalg/interpolative_module with a custom command
      [272/1619] Compiling Fortran object scipy/special/libspecfun.a.p/specfun_specfun.f.o
      [273/1619] Generating 'scipy/special/_ufuncs_cxx.pypy38-pp73-x86_64-linux-gnu.so.p/_ufuncs_cxx.cpp'
      [274/1619] Generating scipy/linalg/fblas_module with a custom command
      [275/1619] Generating 'scipy/special/_ellip_harm_2.pypy38-pp73-x86_64-linux-gnu.so.p/_ellip_harm_2.c'
      [276/1619] Generating 'scipy/special/_test_internal.pypy38-pp73-x86_64-linux-gnu.so.p/_test_internal.c'
      [277/1619] Generating 'scipy/stats/_sobol.pypy38-pp73-x86_64-linux-gnu.so.p/_sobol.c'
      [278/1619] Generating scipy/linalg/flapack_module with a custom command
      [279/1619] Generating 'scipy/special/_ufuncs.pypy38-pp73-x86_64-linux-gnu.so.p/_ufuncs.c'
      [280/1619] Generating 'scipy/special/cython_special.pypy38-pp73-x86_64-linux-gnu.so.p/cython_special.c'
      ninja: build stopped: subcommand failed.
      INFO: autodetecting backend as ninja
      INFO: calculating backend command to run: /usr/bin/ninja
      + meson setup --prefix=/usr /tmp/pip-install-ecigaz4z/scipy_bae3c05728ed48d0a94e36cf7aeb04ac /tmp/pip-install-ecigaz4z/scipy_bae3c05728ed48d0a94e36cf7aeb04ac/.mesonpy-hvxyf1qr/build --native-file=/tmp/pip-install-ecigaz4z/scipy_bae3c05728ed48d0a94e36cf7aeb04ac/.mesonpy-native-file.ini -Ddebug=false -Doptimization=2
      + meson compile
      [end of output]
 
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed
 
× Encountered error while generating package metadata.
╰─> See above for output.
 
note: This is an issue with the package mentioned above, not pip.

解决思路 #

• 升级PyPy版本：当前使用的PyPy 7.3.9是2022年的版本，Ubuntu 24.04软件源可能有更新版本，执行sudo apt update && sudo apt upgrade pypy3尝试升级；若源中版本仍旧，可手动下载安装官方最新稳定版，新版本通常会修复旧版兼容性与段错误问题。

• 安装预编译Scipy包：避免源码编译，执行pip install scipy --only-binary :all:强制安装二进制包；若pip找不到对应PyPy的包，可尝试指定适配旧PyPy的Scipy版本，比如pypy -m pip install scipy==1.10.1。

• 补全编译依赖：Scipy编译除gfortran外，还需libopenblas-dev、liblapack-dev、cython3、meson、ninja-build等依赖，执行sudo apt install libopenblas-dev liblapack-dev cython3 meson ninja-build安装后，重新尝试编译安装。

• 调试出错脚本：进入Scipy临时安装目录（类似/tmp/pip-install-ecigaz4z/scipy_bae3c05728ed48d0a94e36cf7aeb04ac），手动运行出错的脚本：pypy3 scipy/special/utils/makenpz.py --use-timestamp gsl -o scipy/special，用gdb调试段错误：gdb pypy3 --args scipy/special/utils/makenpz.py --use-timestamp gsl -o scipy/special，输入run启动，崩溃后输入bt查看调用栈定位问题。

• 创建干净虚拟环境：当前环境可能存在冲突，创建新PyPy虚拟环境：pypy3 -m venv new_env，激活后source new_env/bin/activate再尝试安装Scipy，排除环境干扰。

内容的提问来源于stack exchange，提问作者cberk1