python的环境，你再也不用愁-conda

Conda Guide

---

Conda简介

conda是一个包，依赖和环境管理工具，适用于多种语言，如: Python, R, Scala, Java, Javascript, C/ C++, FORTRAN。

应用场景：比如在A服务器开发了一个应用，安装了N个包。现在要迁移到B服务器，又要重新安装一遍，还不知道A服务器上哪些包是必须的。conda就是解决这种问题，把该应用需要的包都安装到应用所在的环境中，迁移的时候，只要把环境导出，再导入到B环境即可。

Conda的安装安装过程windows的安装就不演示了，直接在网上搜miniconda安装包，然后一路点下一步即可安装完成。

下边讲解linux下的安装

创建condarc.mirror文件

channels:
  - conda-forge
  - bioconda
  - defaults
show_channel_urls: true
default_channels:
  - /news/upload/ueditor/image/202209/jegcsomjcet
  - /news/upload/ueditor/image/202209/gjxi1mo234m
  - /news/upload/ueditor/image/202209/2pq2wff0lub
custom_channels:
  conda-forge: /news/upload/ueditor/image/202209/ryhdyc3v52t
  msys2: /news/upload/ueditor/image/202209/ryhdyc3v52t
  bioconda: /news/upload/ueditor/image/202209/ryhdyc3v52t
  menpo: /news/upload/ueditor/image/202209/ryhdyc3v52t
  pytorch: /news/upload/ueditor/image/202209/ryhdyc3v52t
  simpleitk: /news/upload/ueditor/image/202209/ryhdyc3v52t

curl -L -o /tmp/miniconda.sh https://repo.anaconda.com/miniconda/Miniconda3-py39_4.11.0-Linux-x86_64.sh
/bin/bash /tmp/miniconda.sh -b -p /opt/conda
rm /tmp/miniconda.sh
conda clean -tipsy
find /opt/conda -follow -type f -name '*.a' -delete
find /opt/conda -follow -type f -name '*.pyc' -delete
conda clean -afy
cp ./condarc.mirror /root/.condarc

更新conda

conda update conda

镜像服务器

conda config --add channels /news/upload/ueditor/image/202209/hjnmfxga4w1
conda config --add channels /news/upload/ueditor/image/202209/joklgv2ym01
conda config --add channels /news/upload/ueditor/image/202209/ou2jzypvik5
conda config --set show_channel_urls yes
# 执行完上述命令后，会在Users目录生成.condarc

环境管理查看所有环境

conda env list

新建环境

conda create --name [name] python_or_others
ps: conda create --name FastAPI python=3.9.12

进入环境

conda activate env_name

退出环境

conda deactivate

删除环境

conda remove -n env_name --all

复制环境

conda create --clone ENVNAME --name NEWENV

package管理列出package

conda list

列出指定环境中的所有软件包

conda list -n myenv

安装package

pip install xxxx 或者 conda install xxxx
ps:pip install tensorflow

如果不用-n指定环境名称，则被安装在当前活跃环境，也可以通过-c指定通过某个channel安装

conda install (-n python34) numpy

更新package

conda update (-n python34) numpy

卸载package

conda remove/uninstall package_name

查找package信息

conda search (-n python34) numpy

更新目前环境所有package

conda update --all

导出当前环境的package信息

conda env export > environment.yaml

清除缓存

删除索引缓存、锁定文件、未使用的缓存包和tarball(压缩包).

conda clean -a

环境的复制注意：yaml的方式，很消耗资源，系统配置至少要2核4G以上，且yaml的package不能过多，否则会被killed1、导出环境

conda env export > environment.yaml

文件内容示例

name: kyle
 
channels:
  - /news/upload/ueditor/image/202209/joklgv2ym01
  - https://repo.anaconda.com/pkgs/main
  - /news/upload/ueditor/image/202209/4b4o01khpop
  - /news/upload/ueditor/image/202209/hjnmfxga4w1/
  - /news/upload/ueditor/image/202209/zigkktmjtjl
  - /news/upload/ueditor/image/202209/o25yg1qlxuj
  - /news/upload/ueditor/image/202209/ou2jzypvik5/
  - /news/upload/ueditor/image/202209/5gx2tz1li3t
  - /news/upload/ueditor/image/202209/joklgv2ym01/
  - defaults
 
dependencies:
  - _pytorch_select=0.2=gpu_0
  - pip:
    - opencv-python==4.1.2.30
 

2、导入环境

conda env create -f environment.yaml

3、Clone环境

conda env update -n my_env --file ENV.yaml