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

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

Loom URL: /items/tracing/new Fabric reference: ai.azure.com — Tracing (span tree + Gantt timeline + per-span input/output + token counts + cost) Loom screenshot: temp/parity/tracing-loom.png

Phase 4

Route Status Notes
GET /api/items/tracing?hours=24 200 Returns 0 traces (App Insights queried but no recent flow runs to trace)

Page shows Window (hrs) input, Operation filter input, Reload button, and an empty table with headers: Time · Operation · Name · Duration (ms) · Success · Result.

Phase 3 — Fabric vs Loom

Fabric element Loom present? Severity
Span tree (hierarchical, expand/collapse parent → child spans for a single trace) NO — Loom is a flat row-per-trace table BLOCKER
Gantt timeline (one bar per span, horizontal axis = wall clock) NO BLOCKER
Per-span detail pane (input / output / tokens / cost / model / latency) NO BLOCKER
Token & cost aggregation per trace + per project NO MAJOR
Filter by status / model / operation / user / tag partial (only operation filter) MINOR
Time-range presets (5m / 1h / 24h / 7d / custom) partial — single hours integer input MINOR
Trace search by content / regex NO MAJOR
Live tail / auto-refresh NO MAJOR
Export to JSON / share NO COSMETIC

Functional

  • Reload button re-calls /api/items/tracing — works
  • Empty state is fine (no flow runs in this hub) but the editor has no example/demo trace to show what it WOULD look like

Grade — D

Backing route is real (App Insights query). UI is a flat KQL-result table — NONE of the span tree or Gantt visualization that defines a "tracing" surface. Per build-phase contract this is BLOCKER for spans-tree visualization. Grade D.

Note: this matches the "Tracing: spans tree with Gantt timeline, or flat KQL output?" critical check in the validator prompt. Answer: flat KQL output.