[Windows] pip install pointnet2_ops报错解决

[Windows]pip install pointnet2_ops报错解决

报错信息👀

(m2t2) PS F:\code_project\Gitub_projects\M2T2-master> pip install pointnet2_ops/ -i https://pypi.douban.com/simple
Looking in indexes: https://pypi.douban.com/simple
Processing f:\code_project\gitub_projects\m2t2-master\pointnet2_ops
  Preparing metadata (setup.py) ... done
Requirement already satisfied: torch>=1.4 in e:\anaconda\envs\m2t2\lib\site-packages (from pointnet2_ops==3.0.0) (2.0.1)
Requirement already satisfied: filelock in e:\anaconda\envs\m2t2\lib\site-packages (from torch>=1.4->pointnet2_ops==3.0.0) (3.13.1)
Requirement already satisfied: typing-extensions in e:\anaconda\envs\m2t2\lib\site-packages (from torch>=1.4->pointnet2_ops==3.0.0) (4.10.0)
Requirement already satisfied: sympy in e:\anaconda\envs\m2t2\lib\site-packages (from torch>=1.4->pointnet2_ops==3.0.0) (1.12)
Requirement already satisfied: networkx in e:\anaconda\envs\m2t2\lib\site-packages (from torch>=1.4->pointnet2_ops==3.0.0) (3.2.1)
Requirement already satisfied: jinja2 in e:\anaconda\envs\m2t2\lib\site-packages (from torch>=1.4->pointnet2_ops==3.0.0) (3.1.3)
Requirement already satisfied: MarkupSafe>=2.0 in e:\anaconda\envs\m2t2\lib\site-packages (from jinja2->torch>=1.4->pointnet2_ops==3.0.0) (2.1.5)
Requirement already satisfied: mpmath>=0.19 in e:\anaconda\envs\m2t2\lib\site-packages (from sympy->torch>=1.4->pointnet2_ops==3.0.0) (1.3.0)
Building wheels for collected packages: pointnet2_ops
  Building wheel for pointnet2_ops (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py bdist_wheel did not run successfully.
  │ exit code: 1
  ╰─> [40 lines of output]
      running bdist_wheel
      running build
      running build_py
      creating build
      creating build\lib.win-amd64-cpython-310
      creating build\lib.win-amd64-cpython-310\pointnet2_ops
      copying pointnet2_ops\pointnet2_modules.py -> build\lib.win-amd64-cpython-310\pointnet2_ops
      copying pointnet2_ops\pointnet2_utils.py -> build\lib.win-amd64-cpython-310\pointnet2_ops
      copying pointnet2_ops\_version.py -> build\lib.win-amd64-cpython-310\pointnet2_ops
      copying pointnet2_ops\__init__.py -> build\lib.win-amd64-cpython-310\pointnet2_ops
      running egg_info
      writing pointnet2_ops.egg-info\PKG-INFO
      writing dependency_links to pointnet2_ops.egg-info\dependency_links.txt
      writing requirements to pointnet2_ops.egg-info\requires.txt
      writing top-level names to pointnet2_ops.egg-info\top_level.txt
      reading manifest file 'pointnet2_ops.egg-info\SOURCES.txt'
      reading manifest template 'MANIFEST.in'
      writing manifest file 'pointnet2_ops.egg-info\SOURCES.txt'
      creating build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src
      creating build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      copying pointnet2_ops\_ext-src\include\ball_query.h -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      copying pointnet2_ops\_ext-src\include\cuda_utils.h -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      copying pointnet2_ops\_ext-src\include\group_points.h -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      copying pointnet2_ops\_ext-src\include\interpolate.h -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      copying pointnet2_ops\_ext-src\include\sampling.h -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      copying pointnet2_ops\_ext-src\include\utils.h -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\include
      creating build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\ball_query.cpp -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\ball_query_gpu.cu -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\bindings.cpp -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\group_points.cpp -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\group_points_gpu.cu -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\interpolate.cpp -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\interpolate_gpu.cu -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\sampling.cpp -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      copying pointnet2_ops\_ext-src\src\sampling_gpu.cu -> build\lib.win-amd64-cpython-310\pointnet2_ops\_ext-src\src
      running build_ext
      E:\Anaconda\envs\m2t2\lib\site-packages\torch\utils\cpp_extension.py:359: UserWarning: Error checking compiler version for cl: [WinError 2] 系统找不到指定的文件。
        warnings.warn(f'Error checking compiler version for {compiler}: {error}')
      error: [WinError 2] 系统找不到指定的文件。
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
  ERROR: Failed building wheel for pointnet2_ops
  Running setup.py clean for pointnet2_ops
Failed to build pointnet2_ops
ERROR: Could not build wheels for pointnet2_ops, which is required to install pyproject.toml-based projects
(m2t2) PS F:\code_project\Gitub_projects\M2T2-master>

关键报错如下，就是说定位不到符合版本的cl文件：

解谜开始😥

先喂给GPT问下：

得到关键信息，我们需要对应版本的VS Build Tools.

此外查找相关博客，参考链接

了解到_MSC_VER是MSVC编译器的内置宏，定义了编译器的版本。

定位到报错信息中的源文件：

可以看出cl.exe的版本要求：

MINIMUM_MSVC_VERSION = (19, 0, 24215)

在本机电脑中找到了VS2017的MSVC版本：

C:\Program Files (x86)\Microsoft Visual Studio\2017\BuildTools\VC\Tools\MSVC

如何对应版本号？🤨

Microsoft Visual C++ - Wikipedia

我尝试根据表中所列的对照关系，定位到了_MSC_VER为1924的一栏，可以看到应该是VS2019的版本，且版本号为16.4.0。

接下来我们去到VS官网，如下图所示，下载链接对应版本号的BuildTools.

验证🤮

安装完成：

找到默认安装路径：C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools

打开目录又傻眼了，翻了半天没找到MSVC文件夹，好像不太一样欸：

应该是没装全，去包管理点击修改：

勾上C++生成工具，然后右侧选择前两个，开始安装。

安装完成，呼~，似乎看到了希望：

尝试点开了其中一个路径：C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Tools\MSVC\14.24.28314\bin\Hostx64\x64

配置环境变量（便于找到cl.exe）⭐

把这个路径加到系统环境变量的PATH中，尝试一手再次执行：

fa fa fa 好消息是没有cl报错了，注意到Unicode，应该是编码问题，继续定位报错行，如下出现的解码格式：

将oem修改为utf-8一试：

依然报错，少了具体报错行🫵🤮

打印输出match对象（匹配到的version号）：

总结梳理⭐⭐⭐

windows编码问题(autorun)

参考博客

终端由于在注册表设置了autorun（历史遗留问题，应该是大一写C遇到终端乱码学的解决方式🫥），每次启动都会设置编码为utf-8.这样会导致visual studio终端启动报错（vcvarsall.bat）。

如下图所示，找到HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Command Processor，将Autorun删除即可。

调用vs检查编码问题

参考博客

在终端输入如下命令：

cmd.exe /c "E:\VS2019\VC\Auxiliary\Build\vcvars64.bat"

至此，无需在cpp_extension.py中修改解码方式。

Visual Studio版本问题

[VS2019+CUDA] fatal error: #error: – C1189: unsupported Microsoft Visual Studio version!_fatal error c1189: #error: – unsupported microsof-CSDN博客

如果VS和CUDA版本不适配，也会遇到问题，我起初电脑上装的有VS2022，VS2019，2022和我安装的CUDA11.1不适配，这里报错图不再贴出。

经过卸载重装，下了VS2019和VS2017，但是依然报错，随后参考了一篇博客内容，修改MSVC的版本控制。

系统CUDA与conda安装的CUDA

注意区分……

ninja包

参考博客

这个错误的出现主要是因为pytorch版本大于1.5，可行的解决方法有三个：

1. 将pytorch版本降到1.5以下；（本末倒置）

2. 将报错的源码torch/utils/cpp_extension.py中的[‘ninja’，‘-v’]修改为[‘ninja’，‘—version’]（不推荐，试过后产生了新的报错……）

   

3. 将pointnet2/setup.py中cmdclass修改，pytorch默认使用ninjia作为backend，这里把它禁用掉。

   

编译成功🫵😁

究极折磨，我为什么要在windows上装🥶🥶🥶