Summary and interpretation of CONDA virtual environment

Preface

csdn There is a lot about conda The article , But there has never been a macro article , I will start from a macro perspective , Integrate the article , Reading , Turn new students from conda Environment entry to configuration , Apply the whole process to interpret .

Of course , Because this article is macro , Maybe many students can't understand it at one time , No problem , This can be seen repeatedly , Help can be provided at different stages .

conda Environmental Overview

conda What's the usage?

You usually need to use conda When ,python Have learned a certain degree , Start some applications , Then there will be a scene ：A The task requires x,y package ,B The task requires z,w package , however y Bao He z Packet collision , You can't do it at the same time AB Mission .

At this time, you need to use two conda Environmental Science , Install separately xy Packages and zw package , The two environments are isolated from each other , They don't influence each other .

Be specific , such as A The code needs tensorflow 1,B The code needs tensorflow 2, This also requires two environments to host .

further , If you feel that a certain environment is bad , Make a deletion directly , Reloading the environment is also easy .

Good to use , As for the specific how in various IDE of use , How to switch , It is roughly as follows , Very convenient , Once configured, you can switch with one click .

conda Environmental level

I said before. ,conda You can create several virtual environments , Your computer can also be installed python Environmental Science ,miniconda Will also bring their own python Environmental Science , What is the relationship between these environments ？

The first is computer python The environment and Miniconda The environment should be of the same level , When you run it on the command line python When ordered , Which one to choose depends on the configuration of environment variables . Generally installed Miniconda, In particular, add path to... Is checked path after , The default environment will become Miniconda Self contained python.

What is the environment variable

We call these two basic environments , stay miniconda Usually called base perhaps root.

After is conda create Create a virtual environment , These are all sub environments . Although it is called sub environment , But in practice base It's not much different from other environments , Just can't delete , So logically, we raise him to a higher level , Put the overall package in base in .

Miniconda install

And Anaconda Comparison of

Many people recommend installing anaconda, But I recommend miniconda.

anaconda Equivalent to a lot of bags +conda Tools , But in fact, few of these bags can be used , It takes up too much space , And reloading is also troublesome .

miniconda Equivalent to a little basic package +conda Tools . Just install the missing package by yourself .

I recommend using miniconda Start from scratch , Your understanding of bags will be deeper .

Pay attention to the version and the source of the installation package

Many people like to install higher versions , But you go to see the industry , They are all stable versions , On the one hand, the high version is unstable , On the other hand, the environment is not suitable . It is prone to package conflicts in the basic environment .

So it is recommended to install 3.7 edition 64 Bit Miniconda.

also , It is recommended to download from the official website , Under the mirror website , There may be problems , There is a friend here who has been in the mood of mirror image all afternoon , The result is to switch back to the official website .

Miniconda download

conda command

conda There are many commands of tools on the Internet .

conda Common commands

It should be noted that , In a virtual environment （ Include base Environmental Science ） When installing the package , As far as possible with conda instead of pip, because conda Better management of package dependencies . When we conda When I can't be safe , Ability to use pip.

Although in conda Used in the environment pip You can also install the package in the virtual environment , But in the end, there will be unknown effects .

conda install package_name

jupyter notebook install + Expand

Installation level

First of all, you need to know ,jupyter notebook It's a IDE, A user only needs one IDE that will do , and IDE You can call various conda A virtual environment , This explanation jupyter notebook The hierarchy of , Should be miniconda Grade , It is at the basic environment level .

So directly in base Run the command to install

jupyter notebook

conda install jupyter

jupyter install

Virtual environment switching plug-in ：nb_conda

Virtual environment alone is not enough , If you want to be in jupyter notebook Switch inside , Need a plug-in .

conda install nb_conda

however , Later, we need to conda Installation in environment ipykernel Talent .

Theme switching plug-in ：jupyter_themes

Change to a fresh theme , Adjust the layout of fonts , It will be very comfortable .

Switching themes

Extensions collection ：nb_extensions

More plug-ins can improve efficiency .

Extension plug-in

conda Environment creation and basic configuration

Generally, installation requires only two points , One is name, One is python edition

If there is no special requirement , General installation 3.7 perhaps 3.8. No addition python Version command , He'll install 3.10, This version of python It's not very useful .

conda create -n env_name python=3.8

Then you can activate conda The environment is further configured

conda activate env_name

install ipykernel, Otherwise, it will not be nb_conda The plug-in recognizes . Many articles are written about installation jupyter notebook, They compare dishes , Not realizing the essence , It just needs one kernel that will do , There is no need to install the whole jupyter.

conda install ipykernel

conda Perfect environment

Basic commands

Come here , stay jupyter notebook You can switch freely in conda The environment , Then there is environment customization , Install whatever bag you want .

conda install package_name

The case that the package cannot be found

occasionally conda Can't find , You can add -c conda-forge Command use extra , With pynput Library example

conda install pynput -c conda-forge

Last, last , No way out , You can use it again pip

pip install package_name

Download too slow

This is mainly because of the source of the software package

conda Source change

Use scenarios ：IDE Use in

jupyter notebook in

Actually, I've finished this before .

vscode&PyCharm

these two items. IDE You can switch the environment in the lower right corner , among conda The environment can also be switched .

Advanced ：Miniconda Directory structure

The first level is Miniconda root directory , There are several folders below ：

1. bin There are binary files
2. scripts It is also an executable script file
3. Lib yes base Library of environment , among site-packages yes conda install The location of
4. envs Is the virtual environment directory .

envs There are several directories under , The directory name corresponds to the virtual environment name

Enter a virtual environment directory ,Lib Under the directory is the library of the virtual environment

among site-packages yes conda install The location of .

This hierarchy is consistent with what we said earlier , although base When used, it is similar to the virtual environment , But in essence base Is above the virtual environment .