If the interrupt edge tests are running while under heavy CPU load (like building wpilib) they are prone to failure since the interrupt thread doesn't have enough time to set up callbacks. The interrupt edge tests now copy the original AsynchronousInterrupt test, which has a 0.5s delay after the interrupt is enabled. Running the new interrupt tests while building allwpilib causes far less failures than the old tests.
That behavior has not been present since PR #4158 was merged more than 2 years ago and imo should not be added back because it was surprising and not consistent with the most common use case of registering a callback permanently.
This makes it easier to define schemas when the type name is non-trivial (e.g., templated structs).
This is breaking for a) custom struct implementations and b) anything calling `wpi::Struct<T>::GetTypeString(info...)` in C++ directly. In both cases, it's a simple translation: For A, rename `GetTypeString()` to `GetTypeName()` and remove the struct: at the beginning, and for B, use `wpi::GetStructTypeString<T>(info...)` instead.
Update() checks/updates the last value and appends only if changed.
GetLastValue() gets the last value.
Also add OutputStream support to Java DataLogWriter.
If one of the *Init() functions takes several multiples of the nominal
loop time, the callbacks after that will run, then increment their
expiration time by the nominal loop time. Since the new expiration time
is still in the past, this will cause the callback to get repeatedly run
in quick succession until its expiration time catches up with the
current time.
This change keeps incrementing the expiration time until it's in the
future, which will avoid repeated runs. This doesn't delay other
callbacks, so they'll get a chance to run once before their expiration
times are corrected.
The other option is correcting all the expiration times at once, which
would starve the other callbacks even longer so that the callback
scheduling returns to a regular cadence sooner. The problem with this
approach is if a previous callback overruns the start of the next
callback, the next callback could potentially never get a chance to run.
