Closed justinsb closed 2 months ago
Hmmm .... this no longer needs a rebase (at least it shouldn't)... I'll try a retest
/retest
Also, wouldn't hurt to cover/document this behavior in tests :)
A great point, I was thinking about that the other day. This change was to support testing in k8s-config-connector, so it got test coverage there, but we could easily create some simple tests directly in this repo - let me do that!
This is escalating a bit ... looks like I need to break out a new module to avoid a circular dependency: https://github.com/kubernetes-sigs/kubebuilder-declarative-pattern/pull/398
Had a bit of trouble with the go.sum / go.mod checking (dev/format-gomod), but I finally figured out that the go.sum difference was to do with my cache; I did go clean -cache -modcache
before running dev/format-gomod
. I think it's still worth trying go.work (and I added a note to the dev/format-gomod script and to the GitHub action error message)
This might be much easier if we merged a (basically empty) ktest
module first - then I think we could stay away from go work sync
which is the root of all the problems. Maybe that's a reasonable compromise that's worth a go.
I tried staging the ktest module in #400. I think that would mean I could back out the go work sync
logic which gave me so much trouble
Found a repo that checks in go.work: kubernetes/kubernetes ! I think we're on solid ground :-)
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: justinsb, yuwenma
The full list of commands accepted by this bot can be found here.
The pull request process is described here
This is an edge case in the kube apiserver, but there is special handling for the stringData field of a secret, that is mapped to base64 data.