pytorch gpu版本安装外部库报错 TypeError: LoadLibrary() argument 1 must be str, not None

记一次靠发懵/直觉/试错解决bug的经历，总之不是完全靠思考···，很有趣

刚刚装完显卡，vs, cuda, cudnn, pytorch gpu版本一串软件····

结果打开以前的程序，发现matplotlib，joblib等等基础的外部包没了，毕竟换了解释器，这是题中应有之义，

但是安装不了，一安装就直接报错！！！这就有毛病了

（问题已经解决了，随便找了个包academic复现之前的错误类型，重点是错误）

把执行命令复制下来在终端执行，看更详细的报错信息：

(base) C:\Users\Administrator\PycharmProjects\EmotionRecog>conda activate pytorch_gpu

(pytorch_gpu) C:\Users\Administrator\PycharmProjects\EmotionRecog>D:\ProgramData\Anaconda3\Scripts\conda.exe install -p D:/ProgramData/Anaconda3/envs/pytorch_gpu academic -y

# >>>>>>>>>>>>>>>>>>>>>> ERROR REPORT <<<<<<<<<<<<<<<<<<<<<<

    Traceback (most recent call last):
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda\exceptions.py", line 1074, in __call__
        return func(*args, **kwargs)
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda\cli\main.py", line 84, in _main
        exit_code = do_call(args, p)
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda\cli\conda_argparse.py", line 82, in do_call
        exit_code = getattr(module, func_name)(args, parser)
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda\cli\main_install.py", line 20, in execute
        install(args, parser, 'install')
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda\cli\install.py", line 116, in install
        if context.use_only_tar_bz2:
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda\base\context.py", line 666, in use_only_tar_bz2
        import conda_package_handling.api
      File "D:\ProgramData\Anaconda3\lib\site-packages\conda_package_handling\api.py", line 3, in <module>
        from libarchive.exception import ArchiveError as _LibarchiveArchiveError
      File "D:\ProgramData\Anaconda3\lib\site-packages\libarchive\__init__.py", line 1, in <module>
        from .entry import ArchiveEntry
      File "D:\ProgramData\Anaconda3\lib\site-packages\libarchive\entry.py", line 6, in <module>
        from . import ffi
      File "D:\ProgramData\Anaconda3\lib\site-packages\libarchive\ffi.py", line 48, in <module>
        libarchive = ctypes.cdll.LoadLibrary(libarchive_path)
      File "D:\ProgramData\Anaconda3\lib\ctypes\__init__.py", line 434, in LoadLibrary
        return self._dlltype(name)
      File "D:\ProgramData\Anaconda3\lib\ctypes\__init__.py", line 356, in __init__
        self._handle = _dlopen(self._name, mode)
    TypeError: LoadLibrary() argument 1 must be str, not None

An unexpected error has occurred. Conda has prepared the above report.

Upload successful.

这个问题简直一点思路都没有，让我十分痛苦，看到网上很多人出现同样的问题，但是别人都只有一个base 解释器，和我的情况不太一样，我装cuda版本的pytorch过程中创建了新的解释器pytorch_gpu，这个pytorch_gpu只是个名字，任意起的哈
大家说的比较可信的一个说法：dll文件的问题，无法动态链接到库，但我还是不知道怎么解决，只是觉得这是对这个错误的正确解释

怎么都没解决，僵持了很久，叹了不少的气，出了不少头皮油之后，看着屏幕上的安装package的界面，觉得conda的绿色圈圈挺好看的，鬼使神差地想着点点它在安装试试，结果···guess what? 这个绿圈圈就是终极答案！！！超级像密室逃脱解密游戏有没有······随便乱试的或者靠着女人神奇的intuition， the sixth sense, whatever,就这么通关了啊哈哈哈哈，棘手bug通关后的喜悦之情溢于言表

解决问题的关键和核心： conda package manager！！！

是否使用conda包管理器左边显示的包都是不一样的。

点击绿圈圈使用conda包管理器之后，瞬间安装成功

后面在pycharm的终端安装了一下上面用来举例的academic包，结果很正常的就装好了，仔细一看，原来是因为终端默认使用base解释器，所以不需要conda包管理器，就可以正常安装外部库

(base) C:\Users\Administrator\PycharmProjects\EmotionRecog>D:\ProgramData\Anaconda3\Scripts\conda.exe install -p D:/ProgramData/Anaconda3/envs/pytorch_gpu academic -y
Collecting package metadata (current_repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: D:\ProgramData\Anaconda3\envs\pytorch_gpu

  added / updated specs:
    - academic


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    academic-0.5.1             |             py_0          12 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    bibtexparser-1.1.0         |             py_0          32 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    chardet-3.0.4              |        py37_1003         184 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    cryptography-2.8           |   py37hb32ad35_0         567 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    idna-2.8                   |        py37_1000         100 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    pyparsing-2.4.5            |             py_0          58 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    pysocks-1.7.1              |           py37_0          27 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    requests-2.22.0            |           py37_1          84 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    toml-0.10.0                |             py_0          16 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    urllib3-1.25.7             |           py37_0         159 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    ------------------------------------------------------------
                                           Total:         1.2 MB

The following NEW packages will be INSTALLED:

  academic           anaconda/cloud/conda-forge/noarch::academic-0.5.1-py_0
  bibtexparser       anaconda/cloud/conda-forge/noarch::bibtexparser-1.1.0-py_0
  chardet            anaconda/cloud/conda-forge/win-64::chardet-3.0.4-py37_1003
  cryptography       anaconda/cloud/conda-forge/win-64::cryptography-2.8-py37hb32ad35_0
  future             anaconda/cloud/conda-forge/win-64::future-0.18.2-py37_0
  idna               anaconda/cloud/conda-forge/win-64::idna-2.8-py37_1000
  pyopenssl          anaconda/cloud/conda-forge/win-64::pyopenssl-19.0.0-py37_0
  pyparsing          anaconda/cloud/conda-forge/noarch::pyparsing-2.4.5-py_0
  pysocks            anaconda/cloud/conda-forge/win-64::pysocks-1.7.1-py37_0
  requests           anaconda/cloud/conda-forge/win-64::requests-2.22.0-py37_1
  toml               anaconda/cloud/conda-forge/noarch::toml-0.10.0-py_0
  urllib3            anaconda/cloud/conda-forge/win-64::urllib3-1.25.7-py37_0
  win_inet_pton      anaconda/cloud/conda-forge/win-64::win_inet_pton-1.1.0-py37_0



Downloading and Extracting Packages
pyparsing-2.4.5      | 58 KB     | ################################################################################################################################################ | 100%
toml-0.10.0          | 16 KB     | ################################################################################################################################################ | 100%
pysocks-1.7.1        | 27 KB     | ################################################################################################################################################ | 100%
chardet-3.0.4        | 184 KB    | ################################################################################################################################################ | 100%
bibtexparser-1.1.0   | 32 KB     | ################################################################################################################################################ | 100%
cryptography-2.8     | 567 KB    | ################################################################################################################################################ | 100%
academic-0.5.1       | 12 KB     | ################################################################################################################################################ | 100%
requests-2.22.0      | 84 KB     | ################################################################################################################################################ | 100%
idna-2.8             | 100 KB    | ################################################################################################################################################ | 100%
urllib3-1.25.7       | 159 KB    | ################################################################################################################################################ | 100%
Preparing transaction: done
Verifying transaction: done
done

而**pytorch_gpu解释器之后，再这么装，就要报文首提的错，这是因为pytorch_gpu解释器需要conda包管理器才可以正常安装外部库

(base) C:\Users\Administrator\PycharmProjects\EmotionRecog>conda activate pytorch_gpu

(pytorch_gpu) C:\Users\Administrator\PycharmProjects\EmotionRecog>D:\ProgramData\Anaconda3\Scripts\conda.exe install -p D:/ProgramData/Anaconda3/envs/pytorch_gpu academic -y

# >>>>>>>>>>>>>>>>>>>>>> ERROR REPORT <<<<<<<<<<<<<<<<<<<<<<

重要结论说三遍：

pytorch_gpu解释器需要conda包管理器才可以正常安装外部库

pytorch_gpu解释器需要conda包管理器才可以正常安装外部库

pytorch_gpu解释器需要conda包管理器才可以正常安装外部库

base解释器不需要conda包管理器，就可以正常安装外部库

base解释器不需要conda包管理器，就可以正常安装外部库

base解释器不需要conda包管理器，就可以正常安装外部库