Rebuild readiness is a gap register, not a promise

Rebuild readiness is a review state, not a generated verdict.

A database inventory may show tables, fields, relationships, saved-query kinds, forms, reports, controls, modules, and macros. That inventory is a starting point. A rebuild plan also needs behavior, ordering, business meaning, external dependencies, permissions, data rules, and acceptance criteria that static extraction may not observe.

TraceMap therefore keeps conclusions attached to rule IDs, evidence tiers, repository-relative spans, commit identity, extractor versions, coverage labels, supporting fact IDs, and limitations. A missing relationship is not silently converted into absence. It becomes an explicit owner question or a Tier 4 gap.

Action queries and crosstabs reveal shape, not results.

legacy.access.query.v1 records bounded saved-query categories. Action-query evidence can distinguish append, make-table, update, delete, bulk, or compound mutation shapes where the protected input supports that category. Crosstab evidence can retain declared row-heading, column-heading, value, aggregate, and bounded pivot-alias provenance. Those categories help an owner identify likely migration work without publishing query text.

Source and target participants or field correspondences remain candidates unless role-specific evidence supports them. TraceMap does not execute the query, inspect rows, calculate runtime pivot columns, or infer transaction boundaries. Concatenated query text, unsupported parameters, external sources, and incomplete query catalogs stay visible as reduced coverage.

A mutation shape is not proof of a copy workflow.

legacy.access.copy-clone-candidate.v1 composes already-persisted query and flow evidence. An append shape may become a bounded bulk-append-shape candidate; a make-table shape may become a table-creation-shape candidate. Update, bulk, and compound shapes remain review-tier mutation candidates. Select, delete, pass-through, unknown, and name-only evidence do not become copy candidates.

Names containing “copy,” “clone,” “duplicate,” or “new” are never evidence by themselves. The report does not infer business intent, source-to-target direction, field mapping, generated-key handling, parent/child sequence, semantic row equivalence, or transactionality. A candidate without a matching bounded flow path remains useful query evidence and gains a flow availability gap.

Bindings and events add behavior candidates with hard limits.

legacy.access.binding.v1 can connect declared record sources, control sources, lookup sources, projection ordinals, and master/child correspondences. legacy.access.event-binding.v1 can connect a populated event property to an exact same-module procedure candidate. Calculated expressions, supported domain-function categories, command controls, and literal conditional lookup assignments can widen the set of questions an owner should review.

None of those candidates establish event firing, branch feasibility, lifecycle order, user navigation, write sequencing, or runtime state. Embedded macros, dynamic dispatch, indirect handlers, unsupported expressions, and concatenated query construction are explicit gaps rather than guessed edges.

Composition preserves cycles, ambiguity, and fan-out.

legacy.access.screen-data-flow.v1 composes bounded paths from UI surfaces through bindings, queries, expressions, events, and persisted fields. Each hop retains its supporting facts and evidence strength. Tier 2 structural declarations can coexist with Tier 3 syntax or textual candidates and Tier 4 unknown gaps. A path cannot be stronger than its weakest required hop.

Cycles are reported rather than traversed forever. Fan-out is capped rather than presented as a complete neighborhood. Ambiguous participants remain alternatives. External links stay categorical and protected. These are useful constraints for scoping owner review, not evidence that a workflow is reachable or complete.

Bounds are part of the evidence contract.

Deterministic ordering makes the same bounded input produce the same public-safe projection. Candidate, path, finding, and gap limits keep a dense legacy application reviewable. When a limit omits rows, the output records truncated state, an omitted count, partial coverage, and a rule-backed TruncatedByLimit gap. It does not treat the retained rows as the complete application.

legacy.access.coverage-gap.v1 carries unavailable, incomplete, unsupported, ambiguous, or protected evidence forward. Macro bodies remain unavailable under legacy.access.macro-gap.v1. An empty candidate set can therefore mean “nothing supported was found within this evidence boundary,” never “the behavior does not exist.”

The local review bundle keeps the gaps beside the evidence.

The documented local Access review bundle collects the Access evidence scope, release-review projection, explorer, manifest, deterministic file hashes, coverage state, limitations, and owner questions in one private directory. Its default finding bound is explicit, and replacement is limited to an intact TraceMap-generated bundle. The bundle remains local by default; it is not a public evidence pack.

That packaging matters because a rebuild discussion can otherwise detach a compelling diagram from its gaps. Reviewers can see which query shapes and UI paths are supported, which catalogs were incomplete, which rows were truncated, and which conclusions the evidence contract forbids.

Turn unknowns into named owner questions.

Before a rebuild or migration decision, ask an Access owner which macros and dynamic handlers matter, which branches are feasible, how generated keys propagate, what mutation ordering is required, which external links are available, and where effective permission evidence is still missing. Ask which calculated fields are business rules, which runtime pivot columns are expected, which validation behavior users depend on, and what acceptance evidence would be sufficient.

The manager proof path helps keep those questions connected to evidence. Use the evidence model, gap guide, static-versus-runtime boundary, capability index, change-review use case, and limitations when handing the register to owners.

Read the three Access stories as one evidence chain.

First, read how Access evidence is acquired without running the application. Next, follow a form-to-field static lineage trail. This article adds the management layer: how to preserve bounded candidates, omitted evidence, and owner questions without turning either companion story into a reconstruction promise.

What the gap register does not establish.

It does not establish runtime execution, query results, production behavior, complete coverage, successful reconstruction, effective permissions, data correctness, selected branches, event firing, release approval, or operational safety. It does not replace an Access owner, migration architect, database owner, testing, runtime observation, or human approval.

The public article includes no database, serialized definition, raw query text, VBA or macro body, source snippet, credential, connection material, machine-local path, private identity, raw analyzer output, or private validation detail. No LLM, embedding, vector, or prompt classification is used to create the evidence.

A loud gap is more useful than a quiet promise.

TraceMap can make rebuild review more concrete by organizing static query shapes, bindings, event candidates, flow paths, provenance, coverage, and limits. The responsible output is not “rebuildable.” It is a deterministic register of supported evidence and unresolved questions that owners can use to plan the next validation step.