Testing showed this wasn't an issue with timing, and allows for more
safety in user code making mistakes. Places where the extra thread
wouldn't help have been kept non threaded, using a new internal API.
I can't find where the actual code is implemented, and I get errors if I
try to link or compile to any of its functions. Even CANJaguar doesn't
use them, nor did the old impl of CANTalon. Plus looking at the API it
makes no sense anyway, since it doesn't do any buffers , so I think its
worth it to remove it.
Removes CAN.h from the JNI header
Allows us to control multithreaded access and spurious wakeups easier.
closes#509
Switches DS to use new waitForData functionality
Adds a few new functions
HAL_GetPort both allocates the DIO and sets the DIO into DIO mode. The DIO setting was overriding the SPI setting on the port, so SPI chip selects were broken if you had more than one. Instead use CreatePortHandleForSPI, which is a special function that only allocates the DIO handle, but doesn't actually set the DIO into DIO mode.
* Dedicated RoboRIO Toolchain, allow Toolchain Path to change
* Add cCompiler Tool to correctly discover RoboRIO GCC on Mac
* Add @333fred requests for GString and ToolChainPath
* Add Toolchain Path option to README
While it technically doesn't matter what the return type of the assignment operator is since it's deleted, assignment operators should return a reference instead of a value.
Also adds Java API for doing the same to Java threads (standard Java Thread.setPriority only works
when Java is run as root, even if the process has the right privileges).
This does a major cleanup on our gradle files, primarily converting all instances of manual dependency downloading to use the correct configuration-based method, which has the advantage of being both less code and more safe.
Updated to gradle 3.2.1. This also moves all of the task graph listeners for dependency setup to use the gradle model, making it both safer and reducing line count.
The binary size is being increased much less then when we originally
tried this. Using static arrays save a pointer indirection and will help
with cache in real time cases.
This makes our APIs more consistent. With optimizations enabled, doubles are just as efficient as floats on ARMv7, so we should take advantage of the extra precision.
