Closed thockin closed 7 months ago
hmm, integration test CI looks unhappy
Yeah, I am looking at it.
TL;DR: it seems to run openapi-gen and then re-run it witrh --verify. I must be misunderstanding it.
We have a separate go.mod file for the integration test directory. I'm assuming the go.mod there needs to be updated as well
Test should pass now.
The integration test at HEAD is B-R-O-K-E-N. Running it man ually leaves uncommitted changes, but the test passes. The test itself runs openapi-gen (which generates a delta) then runs a verify against the fresh delta, which passes. The test could literally never fail (unless it exits non-zero).
I don't know why the delta exists. It seems to be a regression in formatting? See https://github.com/kubernetes/kube-openapi/pull/458/commits/d1213fd88d6d30090194811a5edbb2b6392237e8
If the delta is expected, it's fixed here. If not, someone show me why?
This PR now changes the test to generate code into the temp dir and diff, just like it does for the reports and other files.
/lgtm
/approve /hold for good kube-openapi output in the k/k PR that pulls this in
I can't really fix k/k until this merges, but almost all of these commits came straight out of my k/k tree
/remove-hold
/unhold
/hold cancel
/approve
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: alexzielenski, liggitt, thockin
The full list of commands accepted by this bot can be found here.
The pull request process is described here
This is needed to support Go workspaces in k/k.
This is done as several small commits so each step can be seen. It has the unfortunate effect of not compiling at each step.
@liggitt @alexzielenski @jpbetz