Packages, `pip`, and Virtual Environments

Packages, pip, and Virtual Environments

Core path: 25 minutes

How can a project use Python code that does not ship with Python without turning the whole computer into one giant dependency pile?

Standard library versus third-party package

Some modules come with Python:

import json
import pathlib
import random

Other packages must be installed separately. requests is one example.

That gives us three different actions that beginners often mix together:

CREATE/SELECT ENVIRONMENT
        ↓
INSTALL PACKAGE
        ↓
IMPORT PACKAGE IN CODE

import requests does not install anything. pip install requests does not automatically add an import statement to your program.

Isolate the project with venv

Create a virtual environment in the project root:

python -m venv .venv

A virtual environment gives the project its own Python package location. That means Project A can depend on one set of packages without silently changing Project B.

Depending on the workspace, your editor may activate .venv automatically. If it does not, the common commands are:

Windows PowerShell:  .venv\Scripts\Activate.ps1
macOS/Linux:         source .venv/bin/activate

Verify which Python you are using with the environment tools provided by your workspace before installing anything.

Install with the selected interpreter

Use:

python -m pip install requests

Then Python code can import it:

import requests
print(requests.__version__)

If you install into one interpreter and run the program with another, you can still get ModuleNotFoundError. That failure is not proof that pip is broken. It is evidence that your install environment and execution environment may not match.

Track dependencies

A simple requirements.txt records what another environment needs:

requests

Install from it with:

python -m pip install -r requirements.txt

The file communicates dependencies. It does not contain the packages themselves.

Guided example: diagnose an import that "should work"

You run:

python -m pip install requests

Then another terminal reports:

ModuleNotFoundError: No module named 'requests'

Do not reinstall five times. Check the environment chain:

Which Python ran pip?
        ↓
Which Python runs the program?
        ↓
Are both inside the same .venv?

The useful fix is to align the interpreter and environment, not to guess.

Dependency cost

Every third-party dependency adds maintenance and supply-chain surface. That does not mean "never use packages." It means the dependency should solve a problem worth depending on.