search

kubernetes-sigs / kubebuilder-declarative-pattern

A toolkit for building declarative operators with kubebuilder
Apache License 2.0
251 stars 84 forks source link

WIP: test: use go workspace when testing #394

Open justinsb opened 1 month ago

k8s-ci-robot commented 1 month ago

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: justinsb

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files: - ~~[OWNERS](https://github.com/kubernetes-sigs/kubebuilder-declarative-pattern/blob/master/OWNERS)~~ [justinsb] Approvers can indicate their approval by writing `/approve` in a comment Approvers can cancel approval by writing `/approve cancel` in a comment
tomasaschan commented 3 weeks ago

I think this might be what currently blocks builds on a bunch of PRs; if I do this locally (including adding the submodules) the tests that fail on those PRs start passing again.

Is there a reason to not just check in the workspace file in the repo (and use it "always" and not just in the test runs)?

justinsb commented 3 weeks ago

I think this might be what currently blocks builds on a bunch of PRs; if I do this locally (including adding the submodules) the tests that fail on those PRs start passing again.

So I've re-discovered what I had forgotten, we already do this in hack/ci/test.sh. I've started putting those under dev/presubmits/... (or dev/ci/presubmits/...) so maybe that would be more discoverable.

That said, I think we should also ensure that dev/update-golden either requires a go.work or creates one temporarily, and probably nudge developers to run dev/update-golden when the other tests fail.

Is there a reason to not just check in the workspace file in the repo (and use it "always" and not just in the test runs)?

The only reason is that I haven't seen it done elsewhere. But I agree with you, I'm increasingly wondering when we would not want it. My biggest concern is that it breaks usage as a library somehow, in the same way that go.mod rewrites can sometimes cause problems.

tomasaschan commented 3 weeks ago

I though the whole point of go.work was to enable rewrites and similar tweaks in development while explicitly not affecting usage when importing the module. Maybe worth verifying, but unless I'm completely off here I think it'd be nice to just commit this and ensure the normal workflows with eg plain go test Just Works:tm: