Add unit testing (and rework the sensors a bit)

[DrewBrandt]

Summary:

This PR does a lot of restructuring and somewhat significantly modifies the behavior of the sensors.

I've added numerous (like, 50) unit tests for various parts of the base sensor classes, the state class, and the APRS encoding. These tests are only designed to be used on PlatformIO at the moment, and uses its structure for testing. They rely on the Unity testing framework which, according to Unity, is mostly just a collection of easy-to-run asserts. I don't know enough about how the Arduino IDE handles testing to know if this would compile properly on it. I did try to make sure that the library being used in the Arduino IDE would compile properly, though. Further testing for this is required.

Tests currently include:

• APRS messaging
• GPS functionality
• Barometer functionality
• Matrix
• Circular Buffer

More testing documentation to follow.

Changes:

Major:

• Added a mock Arduino framework for use in testing when using the native, or desktop, environment. Everything for the native testing environment is located in either ./test/ or ./lib. This should allow Arduino IDE to properly compile the library using only ./src/ Arduino functions mocked include:
  • delay()
  • digitalWrite()
  • millis()
  • the Serial class (write, print, println)
  • SdFat (with FsFile and SdFs)
  • HIGH and LOW pin values
  • INPUT and OUTPUT pin modes
• Added fake sensors designed to test the base implementation of each sensor type. Includes an IMU, GPS, and Barometer.
• Added a NativeTestHelper.h that sets up common test variables and included files.
• Added a native environment to the platformio.ini to allow for testing on a desktop.
• Reworked the way sensors should be initialized. Call [SensorType].begin() instead of [SensorType].initialize() Full reasoning will be made clear when more documentation is written.
• Added a circular buffer type to the Math folder. Designed to be used with the sensors for storing calibration data, but can be used by anyone to hold anything. Documentation on how to use will come later.
• Removed all references to imumaths in favor of our own mmfs::vector, with the sole exception of the BNO hardware classes, where a conversion method is included.
• Added conditional includes for many places where <Arduino> is included, replacing it with a condition to possibly include the mock version if necessary.
• Introduced a calibration buffer to Barometer and GPS. These optional buffers consistently average the latest data to keep the "origin" point of their readings to 0, essentially combating drift. Call Sensor.markLiftoff() to save the latest value and stop calibration. More documentation to come later.
• Removed the unused time functions from the GPS. If it needs to be re-implemented, a complete re-write would be best anyways.
• Changed GPS's pos vector of size 2 to a position vector of size 3, adding altitude to the z component.
• Changed GPS's distance calculation away from a flat-earth approximation to a faster approximation of Vincenty's Formulae. Taken from this article and this repo. (Supersedes #29 with better approximation)
• Renamed the ERROR LogType to ERROR_ because apparently Windows decided to make ERROR a macro :/

Minor:

• Moved TestingImplementation folder into its own Examples folder outside of src.
• Fixed library.json to point the repo url to TRT instead of my account (holdover from my fork and initial setup of that file).
• Added Abhi and Avi as authors to library.json.
• Removed KF repo from list of dependencies.
• Removed duplicate Windows function definitions in BlinkBuzz.h.
• Added several more constants designed to help unify time lengths to Constants.h.
• Removed UNKNOWN_ type from SensorType and ActuatorType.
• Added a += operator to the vector class.
• Added more header files to MMFS.h.
• Changed APRS message types to be const, as they are string literals.
• Added a proper double APRS base91 encoder to prevent int rounding or truncation.
• Fixed decodeBase91() call for altitude attempting to decode only 2 of the three digits.
• Sensors can now be fed a char literal for their name in the constructor, with a default to the name of the sensor.
• Hardware level updating now occurs through a read() (not update() anymore) function, where ONLY the hardware reading is done. This function should almost always be called ONLY internally by SensorType.update(). More documentation to come later.
• Slightly modified BMP390 initialization to ensure all commands were properly registered.
• Fixed faulty altitude determination that was mistakenly using measured ground pressure as current sea level pressure.
• Added getASLAlt() and getAGLAlt() to barometer to differentiate between absolute altitude and the delta from power on (or launch) to now.
• Added both ASL and AGL (Above Sea Level and Above Ground Level) fields to the Barometer data string

Major Notes:

The IMU implementation in this version is currently... well... not implemented. Changes to IMU hardware brought up the fact that we need to add math to get global frame from the IMU's local frame. I can do this math, but I feel more discussion needs to happen on wwhat this looks like. I think Ezra would say to not even worry about it because you can just put the raw values into the KF and still get the values you need (at least that's my understanding, but I admittedly don't understand how Kalman Filters work).

The State tests are currently not implemented as the IMU is still not implemented. Once all sensors are done, state tests can begin.

I will write documentation about how to both use and modify this code as I begin migrating the Avionics wiki over here. I think I may slightly change the layout so that the wiki is split into two broad categories: using the software and modifying the software. One goes over the API and the other dives into the internals to help people understand how it currently works so that they can make changes if necessary.

Updates after KF and RecordData updates:

• Changed log types to include _ after the name (ERROR_, etc) because Windows included ERROR as a macro and I couldn't find a good way to undefine that macro.
• Added Arduino.h to SDCardStorage, not sure how it built before without that there.
• Changed the sensor read() and init() functions to protected
• Removed logging data from GPS for now (will become a new issue)
• Changed the BNO055's update() to read() like it's supposed to be

[varun-un]

Should the Math folder be renamed to something like Utils if it now has utilities like the circular buffer

[jhauerst]

Before we merge I think we should have some documentation attached about how to use the unit tests somewhere, and how to setup Unity

If it doesn't build with the Arduino IDE, we may want to wait to merge it, but if the "fake" environment you mentioned works to at least avoid build errors, then it should be fine.

I agree that we should have at least some documentation on how to use the tests before this is merged. It doesn't have to be anything super detailed, just some basic steps for running the tests so that people will know how to use them.

[mallamacimj]

Should the Math folder be renamed to something like Utils if it now has utilities like the circular buffer

IMO we should move the circular buffer to a Utils folder but keep all the math in a math folder. I think that will be the most logical to anyone new to the code.

[DrewBrandt]

https://github.com/Terrapin-Rocket-Team/Multi-Mission-Flight-Software/wiki/Using-Unity-Unit-Testing-Framework

Basic Unity Docs

[DrewBrandt]

Updates after KF and RecordData updates:

• Changed log types to include _ after the name (ERROR_, etc) because Windows included ERROR as a macro and I couldn't find a good way to undefine that macro.
• Added Arduino.h to SDCardStorage, not sure how it built before without that there.
• Changed the sensor read() and init() functions to protected
• Removed logging data from GPS for now (will become a new issue)
• Changed the BNO055's update() to read() like it's supposed to be

[jhauerst]

Docs look good to me and tests ran without issue