Nearly every function can be tested via simple input/output. This means that instead of porting all the tests to each repo, we could put all the test data in a yaml file, and use that for tests.
We should pull the test data out of the tablecloth-rescript test files, as that is the most comprehensive at the moment.
We could also make that what we release: a list of functions that are expected and a list of test cases for them.
For implementations, we could write generators to generate test code from the yaml file.
Nearly every function can be tested via simple input/output. This means that instead of porting all the tests to each repo, we could put all the test data in a yaml file, and use that for tests.
We should pull the test data out of the tablecloth-rescript test files, as that is the most comprehensive at the moment.
We could also make that what we release: a list of functions that are expected and a list of test cases for them.
For implementations, we could write generators to generate test code from the yaml file.