testing

There are many test points already defined in language/src/tests/scala using ScalaTest. In general, any change in language should be done in TDD style with tests cases written before code to make that test case pass. This is how the parser and validator were created. That tradition needs to continue.

In language/src/test/input there are a variety of tests with .check files that have the same basename as the .riddl file. The .check files have the error output that is expected from validating the correctly parsing . riddl file. This way we can identify changes in error and messages. These tests are performed by the CheckMessagesTest.scala test suite which will read all the riddl files in test/input/check and check them for validity and compare the output messages to what’s in the .check file. If there is no corresponding .check file then the .riddl file is expected to validate cleanly with no errors and no warnings.

This is where most regression tests should be added. The input should be whittled down to the smallest set of definitions that cause a problem to occur, and then it should succeed after the regression is fixed.

In examples/src/test/scala there is a CheckExamplesSpec.scala which runs the parser and validator on the examples in examples/src/riddl. Each sub-directory there is a separate example. They are expected to parse and validate cleanly without issue