I ran the benchmark in a tmpfs with an Intel Core i5-2430M. I ran it three times for each combination of build invokation and source tree.
First, I tested "make". For master (eb7d55f), I measured an average of 42.751s with a standard deviation of 0.372s. For this commit, I measured an average of 33.394s with a standard deviation of 0.140s. There was a 9.356s, or 22%, improvement with a total error of 1.3%.
Second, I tested "make -j4". For master (eb7d55f), I measured an average of 21.723s with a standard deviation of 0.158s. For this commit, I measured an average of 16.823s with a standard deviation of 0.340s. There was a 4.900s, or 23%, improvement with a total error of 2.7%.
Change-Id: Idb3adce62ed8ef449360c6583896b6da3565cf58
Loops were converted to their range-based equivalents, variable types were replaced with auto where the type was already specified on the same line, the override keyword was added, and instances of NULL and assignments of 0 to pointers were replaced with nullptr.
Change-Id: If281e46a2e2e1c37f278d56df9915236d4b2c864
The changes made in this commit do not affect any actual code,
they are purely aesthetic. I ran clang-format with google style
over all .h/.cpp files in wpilibc that weren't in wpilibC++Sim
or gtest, and the eclipse formatter over all of the Java files
using the Google eclipse formatting configuration.
Change-Id: I9627bca0bc103c398ecc1c5ba17467193291ae63
Messed with some gyro and CANJaguar tests.
Note: The encoder for the CANJaguar slips a great deal with the motor accelerates rapidly.
Change-Id: I09547f95b122eb48b4f97c0496f970cf1a77c5b2
The gyro class no longer attempts to set a default deadband, but it still
has an optional SetDeadband() method.
The gyro integration tests were modified and still pass consistently.
Change-Id: I08a97b00b98b49b0a3c63306fcc809857523af2b
Analog accumulators now wait for the amount of time a full sample
(including oversampling and averaging) lasts after
AnalogInput::ResetAccumulator() is called, so they don't return
old values after being reset.
This delay should be microseconds long and will only happen
when an accumulator is reset.
A new test is is the C++ TiltPanCameraTest that tests this behavior
with the Gyro class.
Change-Id: I1b3ffdeec187959f95c5e637a6d428c9a4bc2cf4
The gyro sensitivity is fixed, the Accelerometer interface is
now used, and some tolerances are adjusted.
Change-Id: Iac1f3c4fbae3be923bd97692684ff72cd2f623f9
Made a toplevel directory for C++ and C++ tests
Change-Id: I4bc2074a7036ec7fe79568b411637a5bee9eb5b3
Added the C++ testing framework and one test
Change-Id: I1e80a1e16b251a49666820a9d4c8caa025da9785
