Python Dependency Management

⚡ The Incident

I was running code on a shared client system where everything was installed globally due to infrastructure constraints (no Docker, no virtual environments, no Conda).

I had:

protobuf==3.17.3

Then I installed:

pip install google-ads==25.1.0

What happened?

Installing collected packages: protobuf
  Attempting uninstall: protobuf
    Found existing installation: protobuf 3.17.3
    Uninstalling protobuf-3.17.3:
      Successfully uninstalled protobuf-3.17.3
Successfully installed protobuf-5.29.4

Boom. Production code relying on protobuf==3.17.x started failing.

🤔 How Did This Happen?

Python’s pip is designed to resolve dependencies lazily. It will happily uninstall one version of a package and install another to satisfy the needs of a newly added package, even if it breaks something else.

In my case, google-ads==25.1.0 required:

protobuf>=4.25.0,<6.0.0

So it upgraded protobuf to 5.29.4, breaking my older code.

---

📄 How Dependency Management Works in Python

Python does not isolate dependencies by default. Unless you’re using tools like:

• Docker: Isolated containers per app (see GroundPlay setup for Docker examples)
• Conda environments: Project-level Python environments
• venv + pip: Built-in virtual environment and installer combo

You’re essentially operating on the same shared space, where any change can break everything.

Why You Should Use Docker / Conda / venv:

• Different apps can use different versions of the same package
• You can roll back broken installs
• You can test before deployment safely

---

❌ What I Did Wrong

1. Installed globally on a shared machine
2. Did not check which dependencies google-ads==25.1.0 would pull in
3. Did not simulate the install first
4. No backup of previous working state

---

✅ What You Should Do Instead

1. Simulate with --dry-run

Check what will happen before installing:

pip install --dry-run google-ads==25.1.0

2. Check Dependencies with pipdeptree

Install it:

pip install pipdeptree

Use it:

pipdeptree --warn

See what’s depending on what, and where conflicts could arise.

3. Backup Before Changing

pip freeze > requirements-backup.txt

Restore later with:

pip install -r requirements-backup.txt

4. Use pip check After Installation

To ensure nothing’s broken:

pip check

---

🤖 Why Does pip Allow This?

Because Python has no built-in dependency isolation. pip doesn’t know if you’re using protobuf for one script or twenty. It just tries to satisfy the latest install request.

This is why professional-grade setups always involve isolation tools like Docker, Conda, or venv.

---

😜 Final Words

To quote the ancient Linux proverb:

If you’re using Linux and tired of Python dependency hell, just uninstall Python. That’ll solve all your problems.

(Joking. Please don’t. Your system probably depends on it. Like… the whole OS.)

---

Stay safe. Use dry runs. Use isolation. And never trust a fresh pip install in prod.

---

Related: GroundPlay setup shows proper Docker isolation practices for production applications.