Testing

Built-in tests

Some of the functionality is covered by tests implemented in Rust. These tests are run via cargo:

cargo test

Config update tests

A set of tests for testing that changing the running configuration works correctly is implemeted as a Python test script. This requires the Python packages pytest and pycamilladsp to run.

Some tests trigger a config reload by sending SIGHUP to the camilladsp process. This is not available on Windows. It uses the pgrep command for getting the PID of the running camilladsp process, and it assumes that only one camilladsp instance is running.

To run the tests, prepare four valid config files, named “conf1.yml” to “conf4.yml”. Example files are available in testscripts/config_load_test. Place the new config files in that folder.

Start camilladsp in wait mode, with the websocket server listening on port 1234:

camilladsp -w -v -p1234

Now start the tests:

cd testscripts/config_load_test
pytest -v

A complete run takes a couple of minutes.

Benchmarks

There are benchmarks to monitor the performance of some filters. These use the criterion framework. Run them with cargo:

cargo bench

FFT Convolution Kernel Benchmarks

Micro-benchmarks for the complex-multiply kernels are available. They compare scalar against NEON on aarch64 and AVX+FMA on x86_64. The bench feature flag is required to expose the kernel functions:

cargo bench --features bench --bench fftconv_kernels