nugglet

nugglet

Robotics and Computer Vision Researcher @ Robotics Innovation Lab, Gamer, (future) Dog Owner

Basics of Writing Unit Test Cases

1 minute read

Published:

Writing good unit test cases is an important part of any software endeavour. But what constitutes a good unit test?

Manual Test Cases

Reasons to use: automation is too expensive, tests are too complex, better for more created/empathetic test approaches Good test cases are easier to maintain, provides versatility and flexibility.

Unit testing can target: verifying functionality, preventing known bugs reoccurring, documenting code, app security (checking for vulnerabilities)

Contents of test cases:

  • Think of non-specific scenarios that are easy to write and maintain
  • Try to cover edge cases

How to write test cases:

  • Keep it simple and consistent. Establishing a style guide for writing test cases may be useful for teams.
  • Start with a good test scenario. Define the test scope well.
  • Be only as specific as necessary. Test cases that overfit to the problem are inflexible and difficult to maintain. 20% of your tests should cover 80% of your features.
  • Know when to be precise. If the test case is too vague, it becomes ineffective. Depending on the test case’s complexity, more details may be needed and it may take a few iterations to find the suitable accuracy. This is particularly so when handling sensitive data. Highly specific test cases can be used to prevent known bugs.
  • Invest time to explore features to discover new test cases and issues. Try to experience the software from a user perspective.
Sample Test Case Format:
	Title: Check updating username with an empty username field
	Prerequisites:  Be logged in and be in the profile settings page
	Actions:
	    1. Scroll to update username form
	    2. Leave username field empty
	    3. Submit the form
	Expectation: Update failure message is returned

Links and Resources

You May Also Enjoy

Intro to RL: Environments, Agents and Heuristics

17 minute read

Published:

Reinforcement Learning (RL) is a subtopic in Artificial Intelligence, which uses a more organic learning approach. It is vastly different from traditional Machine Learning and Deep Learning, which are dependent on swaths of highly prepared data. In contrast, RL models learn primarily through simulation and experience, much like humans. This article covers the very basic concepts used across RL topics.

Segmentation Tasks

5 minute read

Published:

Segmentation is a popular vision problem that is concerned with analysis and object or scene understanding of the contents of a given input scene. Most commonly, this is used with images, but segmentation can also be performed on a range of datatypes such as videos or 2D and 3D point clouds. The ability of segmentation models to detect objects and identify their location makes these models incredibly useful in projects involving autonomous machines and mapping and localization.

Occupancy Grid Maps

3 minute read

Published:

Occupancy grid maps are a type of representation of an environment, and is commonly used by mobile robots for tasks such as path planning, localization, tracking, and many more. This article is an introduction to occupancy grid maps.

Robot Operating System (ROS) for Beginners

23 minute read

Published:

The Robot Operating System (ROS), in short, is a communication tool used in robotics to facilitate transfer of messages or data in a multi-device system. In this article, I go through the very basics of ROS and how to get started with it.