Skip to content
CSA Loom — the Microsoft Fabric experience for Azure tenants where Fabric isn't yet available: lakehouses, warehouses, notebooks, semantic models, Activator rules, Data Agents, across Commercial, GCC, GCC-High, and DoD IL5

ml-experiment — parity gap (validator v2, 2026-05-26)

Loom URL: /items/ml-experiment/new (the /new pattern is not appropriate here because experiments are identified by name/id, not Cosmos-backed; an experiment must be referenced) Fabric reference: ai.azure.com → Jobs/Experiments — run list, run detail with metrics chart, parameters table, artifacts pane, parallel-coordinates, scatter Loom screenshot: temp/parity/ml-experiment-loom.png

Phase 4

Route Status Notes
GET /api/items/ml-experiment (list) 200 0 jobs, 0 experiments (empty hub)
GET /api/items/ml-experiment/new 404 {"ok":false,"error":"No job or experiment named \"new\""} — Loom UI renders "Load failed" MessageBar

The editor crashes into "Load failed — No job or experiment named 'new'" because the URL pattern /new doesn't map to an existing experiment. The runs tree is empty.

Phase 3 — Fabric vs Loom

Fabric element Loom present? Severity
Runs left rail (status badge per run) YES (but empty here) —
Run detail: Overview / Metrics / Parameters / Outputs / Logs / Snapshot tabs partial — Loom shows flat metrics key/value table only MAJOR
Metric charts (line per metric over steps; multi-run overlay) NO BLOCKER
Parallel coordinates ribbon — button visible but dead button visible (no onClick logic) MAJOR
Scatter ribbon — button visible but dead button visible (no onClick logic) MAJOR
Compare runs (multi-select + diff) NO MAJOR
Register model from a run "Register model" button visible but dead MAJOR
Artifacts file browser NO MAJOR
Log streaming pane NO MAJOR
Compute / environment / Docker info NO MAJOR

Functional

  • "New ml experiment" URL crashes because no experiment named "new" exists
  • Reload button reloads the same /new ID → infinite "Load failed"
  • 3 ribbon actions ("Register model", "Parallel coordinates", "Scatter") are all dead — no click handler in source

Grade — F

/new is fundamentally broken for this slug because experiments are read-only history aggregates. The editor renders "Load failed" and several dead ribbon buttons that LOOK like Fabric features. Grade F (vaporware buttons + broken create flow).

Recommendation: the slug should not be wired as a /new flow at all — should be a list view at /items/ml-experiment (no /new) → click a row → open detail.