Closed jjergus closed 4 years ago
Although I'm not sure how useful this is -- should we maybe just pick a few representative tests instead of maintaining a copy of everything?
Are we at the point where we could instantiate the test runner without a test class, just with a lambda?
If so, I'd like to entirely delete these tests (mostly leaving them up to the fbexpect tests) and entirely delete dirty/ , but we still need to make sure that the CLI exits non-zero if given a failing test
Are we at the point where we could instantiate the test runner without a test class, just with a lambda?
Sorry I don't understand, how is that related? I'm probably misunderstanding the purpose of these tests. I was assuming that we want to have a bunch of random (passing) tests just to exercise different parts of the test framework. We also want a bunch of random failing tests to exercise those parts of the test framework (that's what dirty
is for) but that's a separate issue.
The copy of HSL tests in
hacktest/tests/clean/hsl
was outdated and had type errors. This replaces it with a new copy of all the tests from the HSL repository (no manual changes, except I removed therandom
directory because those tests used a trait in a way hacktest's autoloader is not set up to handle).Although I'm not sure how useful this is -- should we maybe just pick a few representative tests instead of maintaining a copy of everything?