The current test runner is a bash script in luther-derive/testsuite/runtests. In addition to not working on Windows it is brittle and verbose in the wrong ways.
The new test runner should have the following features:
Not depend on the current working directory being the root of the project
Output a success indicator for every successful test, not just a failed indicator for failed tests
Suppress the output of the tests unless the test fail (subject to a "debug" flag)
Optionally run each test under kcov
Run a single test by name specified at the command line
It may be possible to parallelize the tests with this runner, but this is more of a stretch goal at this point.
It may be possible to use this approach to hook the new runner into the project as a "cargo testsuite" command.
The quicli crate is probably a good place to start with the implementation of the application.
Once the new application is implemented then it should be hooked into the Travis ci script at ci/check_test_coverage.
The current test runner is a bash script in luther-derive/testsuite/runtests. In addition to not working on Windows it is brittle and verbose in the wrong ways.
The new test runner should have the following features:
It may be possible to parallelize the tests with this runner, but this is more of a stretch goal at this point.
It may be possible to use this approach to hook the new runner into the project as a "cargo testsuite" command.
The quicli crate is probably a good place to start with the implementation of the application.
Once the new application is implemented then it should be hooked into the Travis ci script at ci/check_test_coverage.