Unit Testing & Other Embedded Software Catalysts

An overview of the material of this course and how it relates to our other courses, quizzes, homework, where to find the pencil sharpener, etc.

Unit testing embedded and system code is a relatively new development. In fact, the tool we use in this course was created at a time when many thought such things couldn’t be done.

The nuts and bolts of Section 2 is to push off lots of yacking at you until later in favor of getting you writing tests as soon as possible. Don’t worry. We will address your inevitable questions soon enough. Please be patient.

We need to get you set up before you tear into writing tests. That means being familiar with the documentation we provide and your options for a development environment. In this lecture we discuss:

1. All the core documentation you’ll want to reference regularly plus a variety of thorough treatments of topics related to this course but outside its immediate scope.

2. Options for development (a readymade web-based sandbox or do-it-yourself local environment).

Note: We are beginning the process of transitioning the course project development environment from web-based SourceLair (discussed in this video lecture) to locally-installed Docker that is discussed in the next lecture. Please choose SourceLair or Docker for developing your project. We apologize for any added complexity due to the new option.

Here we introduce an alternative—and eventual replacement—for SourceLair as your development environment. All the needed tools and project materials are included in a Docker image you can execute locally. Please choose SourceLair or Docker moving forward.

Note: We are beginning the process of transitioning this course from using SourceLair to using a locally executed development environment packaged up as a Docker image. Executing commands at the SourceLair terminal shown in other lectures is virtually identical to doing so at the command line within the Docker environment.

If you chose to use the web-based development sandbox we introduced in the previous lecture, in this lecture we walk you through that environment and how it will be used in the course. Together we will verify that your setup is working properly.

Note: Running commands in SourceLair’s terminal are virtually identical to doing the same using the Docker-based local development environment we introduced as an alternative to SourceLair in the preceding lecture. The big differences between SourceLair and the Docker-based environment are (A) Web-based vs. local execution (B) Using a web-based code editor vs. using your own favorite local code editor.

This lecture covers your options for asking questions not addressed in the course material. For instance, you may need help debugging issues in the previous step (lecture). Or, you may have specific technical or philosophical questions that arise during the rest of the course.

The testing setup we employ early on is easy to use and generally problem free. Our supplemental documents cover lots of issues and potential questions. But if you're stuck, you have several options for help.

In this lecture we guide you through developing and running your very first shiny new unit tests using Unity!

A tiny pop-quiz to verify you understood the last section

In this lecture we guide you through developing a full set of unit tests for a real source code module! Along the way we show you how to think through just what exactly should be tested and then how to structure your unit tests to accomplish this.

We give you a challenge where you can apply the skills you’ve learned on your own.

In this lecture we explain what's happening under the hood of these testing tools and why the tools were designed the way they are. We break down a project's code in diagrams to show how your source and test code live separately but in harmony.

We get into the nitty gritty of how and where you actually run your executable unit tests:

• Locally (native)

• In a simulator

• On target

And we address the scope of testing approaches:

• Unit

• Integration

• System

For more on the fine details of unit, integration, and system testing, please see our supplemental “Extra Goodness” document (document archive attached to Lecture 4): What’s the difference? Unit, Integration, and System Testing.

Lecture Appendix

If you want to really dig in to understand a testing framework like Unity:

• Go ahead and look at Unity’s source code: https://github.com/ThrowTheSwitch/Unity/

• Take a gander at MinUnit—A minimal unit testing framework for C in 3 lines of code: http://www.jera.com/techinfo/jtns/jtn002.html

Section Wrapup and Review

We introduce Section 3 and lay out how you can grow your source code and test code together, in parallel, at the same time.

In this lecture we walk you through creating a new embedded source module from scratch. You will begin building a real GPIO module and unit testing all the necessary register settings as you build up the code from requirements. You’ll be developing the source and the tests together. We identify and demonstrate good practices as we go.

We verify that you have this whole flow-thing down.

We introduce the concepts of Test-Driven Development and Behavior-Driven Development. Surprise! You’ve already been using these concepts. We explain their principles and many benefits… and how they apply to building up a suite of unit tests.

In this lecture we share useful resources on the concepts covered in the previous lecture. Additionally we point you to a variety of studies on and philosophical discussions about the value and tradeoffs of unit testing itself. With the materials of this lecture you will be able to speak intelligently to your colleagues about both the pros and cons of unit testing practice. We also assign some homework.

In this lecture we further develop the embedded GPIO module we began in this section and we pick up the pace a little. We continue demonstrating test-first and behavior-driven test development for additional pieces of the GPIO interface.

And, we specifically address good practices for working with:

• Newly implemented but mostly blank functions

• Error codes and conditions

• Bounds checking

Registers are fairly simple—except when they’re not. Different processors use different styles of register interaction and produce a variety of behaviors. See our supplemental document (attached in Lecture 4): Testing Special Cases for more perspective on testing tricky register interactions.

We assign some homework for you in the GPIO module of this section to complete using your new skills.

In this lecture we take a step back and discuss two important concepts underlying the decisions we've been making along the way:

• Do the Simplest Thing That Could Possibly Work

• Ya Ain’t Gonna Need It

These are powerful concepts that may feel a little wrong at first. We explain why they are a good way to work and how testing plays nice with these approaches. We then demonstrate them in action in testing the Init() function of our GPIO module.

We give another brief quiz to verify our progress.

In this lecture we work through a case that seems complicated on its surface. Our mantra of Simplest Thing That Could Possibly Work seems to lead us to call source functions in our unit tests that are outside the test case under development. We show how to think through decomposing these rather common cases into tests and also introduce setUp() and tearDown() within the testing framework.

We assign the final homework assignment for this section. Once completed, you will have a fully working embedded GPIO module and a full suite of unit tests to verify that it works.

Section Wrapup & Review

We discuss where all our work so far belongs in the grand scheme of things and explain how and where our testing approach applies beyond only the work we've done so far.

This section will help you think about the “edges” of your code and how testing applies to common scenarios of state machines, math functions, and data containers.

For an in-depth look at testing special cases (e.g. main(), for() loops, various register setting scenarios), see our supplemental “Extra Goodness” document (in the archive attached to Lecture 4): Testing Special Cases.

In this lecture we walk through the development and testing of a State Machine module used to implement a protocol parser. We discuss and demonstrate how all our recently developed skills apply quite well to this sort of module.

You complete the protocol parsing state machine module from the previous lecture using everything you’ve learned.

We walk you through the development of Math Function module with our approach by demonstrating the development of a well-tested low pass filter.

You complete the Fibonacci module from the previous lecture on your own using all that you’ve learned.

We walk through the development of a data container module (a circular buffer) using all our newly acquired methods and skills.

Once again you complete the circular buffer implementation from the previous lecture using everything you’ve learned along the way.

We review all the work we’ve done in all the sections of our course and where it falls in the scheme of things. Go on towards world domination with your mad scientist plans.

For an in-depth look at testing special cases (e.g. main(), for() loops, various register setting scenarios), see our supplemental “Extra Goodness” document (in the archive attached to Lecture 4): Testing Special Cases. Heck, take a look at all the documents in the archive as they will address many issues we can’t cover in the main course.