Python Testing, Async Requests, and Tooling Updates

PyTest 4.4.0 Release

Brian discusses the latest updates to PyTest version 4.4.0, highlighting several features useful for developers improving their testing workflows:

• Test paths: A configuration setting that allows you to specify exactly where PyTest should look for tests, reducing overhead on large projects.
• Subtests plugin: A new capability that allows tests to continue even after a single subtest failure, and a new subtests fixture that works outside of unittest.
• Async warnings: PyTest will now issue warnings if tests marked as async are not executed due to missing plugins like pytest-asyncio or pytest-trio.

Async Requests and API Mocking

Michael reviews options for working with async/await syntax when making web requests, covering both current workarounds and future directions:

• request-async: A library that provides an async-friendly interface for the popular requests package without requiring a total rewrite of existing code.
• Future-proofing: Discussion on the eventual development of request3, which will natively support async operations, though this is likely a long-term project.
• API Mocking: Tips on using local Flask or Starlette apps as mocks to simulate server behavior during tests, avoiding real network calls.

Community News: NPM and PyPI as a Service

The hosts discuss the complexities of running package repositories as a service, specifically referencing the PyPA ecosystem:

• Financial Hurdles: Running a service like PyPI is expensive, and turning it into a commercial "as-a-service" product could threaten the Python Software Foundation's nonprofit status and donor support.
• Marketplace and Alternatives: Mention of the DigitalOcean marketplace for one-click app deployment and the potential gap for internal package hosting solutions.

Jupyter Notebooks and Tooling

A deep dive into working with Jupyter notebooks in the cloud and educational resources:

• Cloud Platforms: Discussion of services like Binder, Google Colab, Azure Notebooks, and Datalore for running interactive code without local installation.
• Tutorials: Recommendation of DataQuest tutorials for both beginners and advanced users looking to master Jupyter magics and data science workflows.

Python Programming Patterns: Sentinel Values

"Instead of setting a value to none, you can set it to object... it's like the perfect sentinel value."

Michael and Brian explore the pattern of using plain object() as a unique sentinel value, which avoids the common pitfalls associated with using None as a default when None might be a valid input in the data set.

Developer Productivity Tools

• PyTest-neo: A fun plugin that visualizes test execution in the terminal in a unique, descending layout.
• Ace Jump: A powerful productivity tool for PyCharm or IntelliJ that allows developers to quickly jump to any string or code location using keyboard shortcuts.