Krishnan Anantheswaran

Seems to me that the linter error is spurious since the code doesn't compile with that change (given 1.21 in go.mod)

> what matters is the version of the go binary not able to compile locally even with go 1.22.3 ``` $ git diff diff --git a/cmd/crank/beta/trace/internal/resource/xrm/loader.go b/cmd/crank/beta/trace/internal/resource/xrm/loader.go index d6686f40..795a6d27 100644...

Second commit fixes test structure and sorts children after load. Also sets QPS and burst but I will get rid of this with a rebase from master.

Not sure why the E2E tests are failing - the failing test cases seem unrelated to the changes in this PR.

hello, how can I help moving this PR forward? Please let me know.

Benchmarks for 3 cases, before QPS change, after QPS change and current PR. Note that the QPS changes do **not** affect performance since access to resources are sequential with current...

Those are interesting findings and certainly way different from what I have. I've noticed in the past that the network round trip time massively affects performance. One a kind cluster...

Yes, I can reproduce results similar to yours for a kind cluster with ~480 resources 55s drops to 13s after the qps fix and marginal to no improvements for concurrency...

Thanks for the detailed PR review @jbw976 - I resolved all but one of your comments, the code is much better for it!