11 sept. 2026 18:13:49 Arthur Diniz <[email protected]>:
I wanted to flag that several of the failures you listed, especially the ones going through kube-openapi and structured-merge-diff, line up with work I'm currently doing.

I'm in the middle of bumping the Kubernetes-related packages to
bring them up to the 0.36.4 line, which involves the structured-merge-diff v6 switch and the corresponding kube-openapi/gnostic-models updates.

That explains errors like:

cannot use typeSchema.Types (variable of type []"sigs.k8s.io/structured-merge-diff/v6/schema".TypeDef) as []"sigs.k8s.io/structured-merge-diff/schema".TypeDef

and the kube-openapi document_v3.go yaml.v3 type mismatch, these are
transitional and resolve as the dependency chain gets updated together.

Progress so far, already uploaded:

- golang-k8s-sigs-structured-merge-diff
- golang-k8s-klog
- golang-k8s-sigs-yaml
- golang-k8s-sigs-json
- golang-github-google-gnostic-models
- golang-k8s-kube-openapi

Next in the queue to review and upload (in dependency order):

1. golang-k8s-apimachinery -- OK, MR ready
2. golang-k8s-api -- in review
3. golang-k8s-client-go -- in review
4. golang-k8s-cli-runtime -- in review

Until this whole chain lands, expect reverse-dependency autopkgtests to keep failing (client-go, component-base, cli-runtime, kubernetes, cosign, etc. from your list are all downstream of this). Those should clear once the packages above are all in.

While I'm at it, I'm also fixing the packages that currently FTBFS with Go 1.27 and closing the bugs filed against them (for example the encoding/json/v2 default-on change in 1.27 breaks a few test suites).

So I'd treat the kube-openapi / structured-merge-diff / apimachinery-related failures in your list as expected-in-transition rather than new regressions.

Thank you for the clarification, and also for all the work you are doing on this transition!

--
Nicolas Peugnet

Attachment: signature.asc
Description: PGP signature

Reply via email to