fix: use correct JSON Patch add/remove/replace ops in updateQueue

[abhishekKokadwar]

Summary

Fixes #366

updateQueue built every JSON Patch operation as op: "replace", regardless of whether the target field currently exists on the live queue. This caused two distinct problems:

1. Adding a new field fails. Per RFC 6902, replace requires the target path to already exist. Adding a spec field (e.g. guarantee) to a queue that never had one previously caused the Kubernetes API server to reject the whole patch, even though client-side validation had already passed.
2. Clearing a field is silently ignored. Fields the user cleared in the edit form are simply absent from the submitted spec object (see buildSpecFromForm in queue-form-shared.tsx), so no patch operation was generated for them at all — they silently remained on the live queue, contradicting what the UI implied happened.

Fix

Diff the current live queue spec against the submitted spec:

• fields new to the submitted spec → op: "add"
• fields present in both → op: "replace"
• fields present on the current spec but dropped from the submitted spec → op: "remove"

The existing getClusterCustomObject call (previously used only to check the queue exists before patching) is reused to fetch the current spec, so this doesn't add an extra API round trip.

A caveat worth flagging

The queue edit form's Form mode only represents weight/priority/reclaimable/guarantee/capability/deserved. If a queue had some other spec field set outside that set (e.g. via a prior raw YAML edit), saving via Form mode after this fix would now treat that field as removed, since it would be absent from the submitted spec. YAML mode is unaffected since it round-trips the full spec object. I believe this is an acceptable, narrow edge case — arguably it's already broken today in a different way, since Form mode has never had any way to preserve or edit such a field — but flagging it explicitly rather than leaving it implicit. Happy to add a guard if maintainers would prefer one.

Test plan

• npx tsc --noEmit — no type errors
• npm run lint — no new warnings introduced
• npm run build — succeeds, /api/trpc/[trpc] route compiles clean
• Traced the fix against RFC 6902 JSON Patch semantics and against both Form-mode and YAML-mode paths in queue-edit-dialog.tsx
• Not yet verified against a live cluster. I attempted to set up a local kind cluster with Volcano installed to test both scenarios end-to-end (create a queue without guarantee, add one via edit; clear a field and confirm removal via kubectl get queue -o yaml), but the test machine's available memory wasn't sufficient to run kind + Volcano's controllers reliably, and I didn't want to submit a false claim of live verification. If a maintainer can confirm this against a real cluster, or would like me to attempt it again with more resources, I'm happy to follow up.

Diff is scoped to the single affected mutation (23 insertions, 3 deletions).

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[volcano-sh-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign jessestutler for approval. For more information see the Kubernetes Code Review Process.

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

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[abhishekKokadwar]

C:/Program Files/Git/assign @JesseStutler

Hi JesseStutler, this has been open a few days with green CI and DCO signed — would appreciate a review when you have a chance. Happy to make any changes needed.

[abhishekKokadwar]

Note: my earlier /assign comment on this PR was mangled by a local shell path-expansion issue and didn't register as a real command — apologies for the noise. This is still open after ~3 weeks with green CI and DCO signed. Would appreciate a review whenever you have bandwidth.