Interpreter, Virtual Environments, and Dependencies

Objective

By the end of this lesson, you should be able to identify which Python Windows 11 and PowerShell 7 are using, explain the roles of the interpreter, venv, pip, dependency declarations, and lock files, and move from venv + pip to uv when it is useful.

First distinguish five things

Thing	Question it answers	Common check
Python interpreter	Which program executes a .py file?	python --version, Get-Command python
Virtual environment	Which isolated interpreter and packages does this project use?	python -c "import sys; print(sys.executable)"
pip package installer	Which distribution is installed into the current environment?	python -m pip --version
Dependency declaration	Which packages and versions may or must the project use?	pyproject.toml or a requirements file
Locked resolution	Which exact versions did this reproducible installation resolve?	A lock file and its tool documentation

Activating a virtual environment primarily puts its directory at the front of the current shell’s PATH. The activation script also sets VIRTUAL_ENV and may change the prompt. It does not modify source code or guarantee that dependencies have been synchronized from the project declaration.

Version baseline

As of 2026-07-14, Python.org listed Python 3.14.6 as the latest stable release; Python 3.15 was still a prerelease branch. The course examples avoid 3.14-only syntax so they can run in still-supported Python 3.11+ environments. A real project should state its supported range in pyproject.toml or its runbook and verify it in CI and tests instead of assuming that every Python 3 release behaves the same.

The original Python 100 Days course spans several ecosystem eras. Keep the teaching context when you encounter old installation screenshots, package names, or APIs, but use the project’s locked versions and current official documentation as the operational source of truth.

Build a minimum environment with venv + pip

In a new practice directory outside this vault, run:

New-Item -ItemType Directory -Path "$HOME\python-agent-lab" -Force
Set-Location "$HOME\python-agent-lab"
 
python --version
Get-Command python
python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -c "import sys; print(sys.executable)"
python -m pip --version
python -m pip install --upgrade pip

If execution policy prevents activation, directly invoke the environment’s interpreter before weakening a machine-wide policy:

.\.venv\Scripts\python.exe -m pip --version
.\.venv\Scripts\python.exe .\main.py

.venv is reproducible local state and belongs neither in Git nor in this vault. Removing it does not remove project source code, but you need a dependency declaration and lock information before installing again.

Why use python -m pip

A machine can have multiple Pythons and pip.exe programs. python -m pip explicitly means “have this interpreter execute its pip module,” which makes the target environment easier to verify than a bare pip command.

Check all three of these:

python -c "import sys; print(sys.executable)"
python -m pip --version
python -m pip check

The paths should point to the same .venv. pip check verifies whether installed packages declare compatible dependencies, but it cannot prove correct application behavior or supply-chain safety.

Dependency declaration and locking are different jobs

pyproject.toml can declare project metadata, the Python version, and direct dependencies. A requirements file is often used to record or constrain an install set. A lock file is normally produced by a specific tool and records resolved exact versions and the full dependency graph.

A minimal conceptual example:

[project]
name = "agent-lab"
version = "0.1.0"
requires-python = ">=3.11,<3.15"
dependencies = []

The examples in this course use only the standard library, so no third-party dependency needs to be installed. If a project adds an SDK, record why it is needed, its permitted version range, update process, license and security review, regression tests, and rollback target.

Do not put real tokens in dependency files, command history, or installation URLs. Extra package sources can create dependency-confusion risk and must be explicitly configured and reviewed by the organization.

When to use uv

uv is Astral’s third-party Python project and package-management tool. It can manage Python, virtual environments, dependencies, lock files, and tool execution. Understand the preceding objects first, then consider:

uv init agent-lab
Set-Location .\agent-lab
uv run python --version
uv add httpx
uv lock
uv sync

Commands evolve with uv releases, so check the official documentation and lock down team conventions when implementing it. The provenance of Python distributions managed by uv also differs from the python.org installer; verify it where compliance or runtime-origin requirements apply. uv is not a magic layer that removes the need to understand environments.

Common mistakes and diagnostics

Symptom	Check first	Common cause
Installation succeeded but import fails	sys.executable and pip --version	The package was installed into another interpreter
The environment disappears after closing a terminal	Whether it was activated again	Activation affects only the current shell
A teammate cannot reproduce the setup	Dependency declaration, lock file, Python range	Only .venv or an informal version was shared
Behavior changes after an upgrade	Resolved versions and changelog	An unbounded dependency was upgraded without regression tests
Git shows many environment files	.gitignore and status	.venv was accidentally versioned

Exercise

Create an environment outside the vault and record python --version, sys.executable, and python -m pip --version. Close and reopen PowerShell, then run the same script once by activation and once by directly calling .venv\Scripts\python.exe; explain the difference.

Self-check

1. What do the interpreter, virtual environment, and pip each do?
2. Why should .venv not be committed while a dependency declaration should be?
3. Do a version range and a lock file solve the same problem?
4. Why is python -m pip easier to debug than bare pip?
5. Which foundational objects must you still understand before using uv?

Related concepts and next step

• For PowerShell paths and commands, see the cross-platform material in Linux Commands; for version control, see Git.
• Next, Type Hints and Data Models uses explicit types to express tool inputs and outputs.

References

Retrieved on 2026-07-14.

• Python downloads and release status
• Python venv
• Python Packaging User Guide: pip and virtual environments
• Python Packaging User Guide: pyproject.toml
• uv official getting started guide
• uv: managed Python distributions