Hydra-plutus tests are actually red and golden tests got side-stepped
Closed this issue · 2 comments
ch1bo commented
Context & versions
On current master cb5a8c8, we have a green CI https://github.com/input-output-hk/hydra/actions/runs/9280096242.
BUT, the hydra-plutus tests are actually RED.
Steps to reproduce
- Checkout
cb5a8c8e9ed0af7a2f8ba422ab78f042d4ee4920 cd hydra-plutuscabal run hydra-plutus:exe:tests
This is related to #1449 as to verify this, you need to run the tests in the right directory. And a false sense of security is created by seeing the tests green (although they generated new golden files).
Actual behavior
Our CI allowed a change to hydra-plutus to be merged to master without
- forcing to regenerate golden files
- encouraging a CHANGELOG entry that captures this, leading to downstream surprise of #1457
Expected behavior
The CI to actually run the hydra-plutus tests, detect breaking changes by seeing golden tests red (and their intended consequence of documenting detected breaking changes)
Hypothesis
- This Github action run should have caught this https://github.com/input-output-hk/hydra/actions/runs/9256084766/job/25461316118
- But the
nix build .#checks.x86_64-linux.hydra-plutusseemingly does not run thehydra-plutustests.
locallycompact commented