Tutorial: solution review & impact analysis

Persona: solution architect / technical lead · Time: ~20 minutes · Requires: any AOT root (VM or exported metadata).

Prerequisites

• The root open and its index converging — for a full-system review, wait for index-tier-complete (the status bar tells you); partial results are fine for browsing, not for conclusions.

1. Open the root and read the map

Point the Studio at the metadata folder. Within seconds the explorer is browsable — start navigating while deeper tiers parse in the background. The status bar tells you exactly how complete the index is.

2. Answer “who uses this?”

Select any element — a table, a class, a method — and run find-references. The cross-reference web answers across the entire system

Find-references results for SalesTable listing 92 cross-references across the full estate's model tree

: every caller, every reference, through every customization layer. This is the impact-analysis question that otherwise costs an afternoon.

3. Read the code, not the XML

Open any element: X++ renders with full language smarts; Ax*.xml metadata stays navigable — go-to-definition works inside it (index-backed). Jump from a reference to a definition to its callers in one search.

4. Review against the ecosystem’s rules

The best-practice advisor checks the ecosystem’s coding rules continuously. Open the Problems panel for the model you’re reviewing: model diagnostics are separated from collapsed environment noise, so your review reads your code, not Microsoft’s references.

PROBLEMS PANEL · REVIEW VIEW Real UI capture pending (shares capture with xpp-bp-problems) tut-sr-problems-panel

5. Scope a change safely

Before committing to a change: search the source (FTS5, BM25, model-scoped) for the patterns you’ll touch; check the git panel for the root’s working-tree state; and note that search results are flagged partial until the FTS tail converges — for a full-system review, wait for index-tier-complete.

What this replaces

The Microsoft cross-reference rebuild runs on its schedule; this index updates the affected records on each save. For a review cadence — weekly governance, pre-release impact checks — that difference is the whole point.

Verify it worked

• A picked element’s references list every caller you expected — and at least one you didn’t (that’s the review working).
• The Problems panel shows only your model’s diagnostics with the Environment group collapsed.

If it goes wrong

• A negative search surprises you — check the partial flag; wait for convergence before writing the review.
• Pickers feel thin — a lower index tier; give tier 3 time to parse Microsoft models.