Byte-sized RSE: Session 3

Testing your Python Code

Session Introduction

​

Jeremy Cohen, EPSRC RSE Fellow, Imperial College London

​

Steve Crouch, SSI Research Software Group Leader, University of Southampton

The UNIVERSE-HPC (EP/W035731/1) project is funded through the ExCALIBUR programme and is a collaboration between the universities of Edinburgh, Oxford, Southampton and Imperial College London.

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Session Overview

• Testing your code - introduction
• Types of testing
• Mocking
• Related best practices
• Code style/linting
• Continuous Integration (CI)
• Interactive tutorial - unit testing in practice

Byte-sized RSE

Testing your (Python) code

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Testing your code - Why test your code?

• Testing your code enables you to check that it behaves as expected
• Important for supporting maintenance/long-term maintainability
• Can pick up unexpected changes in behaviour of existing code resulting from future changes
• Gives confidence to your code’s users of its correctness and provides invocation examples
• As a codebase gets larger, debugging can become increasingly difficult
• Tests can help you pick up problems before they become a runtime bug
• A failing test will also help pinpoint where the problem lies

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Testing your code

• General principles of testing apply across languages
• Most programming languages provide tools and frameworks to support testing
• A set of tests for a codebase are known as a “test suite”
• Code coverage
• Coverage tools tell you how much of your code is being tested
• Useful to gain insight into which parts of code aren’t being tested

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Testing your code

​

​

A passing test suite (even with 100% code coverage)

DOES NOT MEAN that your code is bug free!!

​

(Writing tests can also be very time consuming - your initial focus should be on prioritising the most critical parts of the code - those that deliver greatest value or involve the most complexity.)

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Testing your code

​

​

Time for a quick quiz!...

​

Go to menti.com and enter code 1762 1286

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Types of testing - Levels

• Unit testing
• Test small, independent pieces of functionality, e.g. specific functions in the code
• Integration testing
• Next level up from unit testing - test blocks of code/modules together to verify they work as expected
• System testing
• Test full system functionality, e.g. top-level inputs/outputs/interactions

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Types of testing - Approaches

• Different approaches exist for the operation of tests
• Where expected results are known, it is possible to check they are provided - either as fixed results or within some confidence interval
• Where exact outputs not known (e.g. a Monte Carlo simulation), property-based testing may be used to run a test with a range of input values that are expected to have a shared output property

​

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

• Things that worked before can stop working after a change to the code - regression testing can be used to identify these “software regressions:

Mocking

• When running tests, you often want to test a very specific piece of functionality
• Dependencies on external objects/functions complicate this (do you know they work as expected?)
• Mocking allows you to replace dependencies with “mocked” objects or functions that behave as you tell them to
• E.g. a function modifies data and writes it to a file
• Mock the file writer object - mocked object stores “written” data
• Check that the data written is as expected, no file created

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Related best practices

• Code style / linting
• Different languages have different code style conventions (e.g. PEP8 in Python)
• Following these makes code more readable by others and easier to maintain
• Linting tools check that code conforms to specified styling properties - often run automatically
• Continuous Integration (CI)
• Automate the running of key processes in response to code changes
• E.g. Automatically run tests and linter on every change/commit

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code

Testing in Python, in practice

• That’s all for the background material
• Now it’s your chance to have a go at writing some tests….

​

​

…over to Steve for the interactive part of the session

The content of these slides is licensed under CC-BY 4.0 © Jeremy Cohen, Imperial College London, Steve Crouch, University of Southampton.

Byte-sized RSE

Testing your (Python) code