Program hood and add DependencyOrderedExecutor to schedule periodic actions

[aidnem]

Adds:

• Hood subsystem with hood constants, test modes, etc
• DependencyOrderedExecutor to schedule tasks in periodic
• DirectedAcyclicGraph implementation to support DOE
• clamps to shot calculator to more accurately reflect capabilities
• Shot calculator sends goal angles to turret and hood and reads actual position from turret and hood for shot projection

Closes #33

[godmar]

This is not how you implement Kahn's. Your implementation is $O(n^2)$ if I see this correctly. It can be done in $O(n + m)$.
Basically, the graph representation should use adjacency lists in the forward direction. That is, if $a \rightarrow b$ and $a \rightarrow c$ are dependencies, you'd have $a \rightarrow [b, c]$. This will allow you to implement the part where a node's successors are found using a simple loop over the successors.

For the incoming part, like I said, don't actually remove edges. Simply keep a count incomingCount in a Map that you initialize with the number of incoming edges (can be done in $O(n + m)$ ), and then decrement the count until it reaches zero. This incomingCount map will be temporary to the routine that computes the top sort.

[aidnem]

I've reimplemented it as you described. Thanks for letting me know.

[aidnem]

Verification:

Here's the mechanism position (blue) following the reference setpoint (gold) as it follows my commanded input (tan).

[godmar]

I just made a brief run through, will look at it more closely later.

[aidnem]

I have fixes for all comments done but my wifi is out so I cannot push changes until it comes back.

[aidnem]

@godmar would you mind looking over some of the architecture changes I made and seeing if you find them to be cleaner than before?

[godmar]

@godmar would you mind looking over some of the architecture changes I made and seeing if you find them to be cleaner than before?

I'm looking at it. Who is going to resolve all these conversation above? I understood that it would be your responsibility?

[godmar]

please check the math in controlToGoalHeading

[godmar]

Side note: with the new test mode manager where each subsystem has its own, the isSubsystemInTestMode functions could be simplified by adding a method to the TestModeManager that returns whether it's in the None state or not (all others would then be presumed to be testing states).

(Implemented here in Theta's branch.)

[aidnem]

Who is going to resolve all these conversation above? I understood that it would be your responsibility?
Yes, sorry I missed a couple when I was resolving them. I left the ones where you asked questions open so that you could resolve if satisfied by the answers or respond if you wanted more clarification/a different decision.

I have to add field locations and then fix the turret math. Those will be my next 2 tasks.

[aidnem]

As a little extra verification of our dependency ordered executor:

Here is the graph with all subsystems enabled:

Here are some graphs after disabling 1 subsystem at a time:

• No homing switch:

  

• No turret:

  

• No hood:

  

Now missing 2 systems:

• No homing switch nor turret:

  

• No homing switch nor hood:

• Homing switch but no turret nor hood:

  

All of these seem to make sense to me. Please let me know if you think any of them are wrong.