Wire harness revisions rarely arrive as clean edits to one stable model. A baseline may be KBL, the current revision may be VEC or 3DXML plus a wirelist, and the released record may be a PDF. File-level diff tools cannot answer the questions the DRE actually owns: what changed, where is it, what does it affect, and is the result safe to release?
Entity matching comes before change detection
The same connector, wire, segment, or protection can have a new identifier, a different source path, or slightly shifted geometry. Comparing raw IDs creates false additions and deletions. Matching only by name misses duplicates. Matching only by position fails when layout changes.
A stronger matcher combines stable identity, part and role, connectivity, neighboring topology, geometry, source provenance, and selected engineering attributes. Ambiguous candidates should remain visible for review rather than being silently paired.
| Comparison layer | Examples | Review question |
|---|---|---|
| Identity | Renamed connector, regenerated ID, changed part occurrence | Is this the same engineering entity or a replacement? |
| Electrical | Endpoint, cavity, gauge, color, signal, splice membership | What circuits and interfaces are affected? |
| Physical | Segment length, branch location, fixing, protection, route | Where does the change occur on the harness? |
| Configuration | Variant inclusion, effectivity, option expression | Which buildable configurations changed? |
| Compliance | New DRC finding, resolved waiver, rule-pack version | Did the revision change release readiness? |
| Documentary | Note, label, dimension, source evidence | Is manufacturing or review intent different? |
Use synchronized physical context
A list of changed fields is necessary but inefficient on its own. Engineers need to select a delta and see the corresponding entity in baseline and current formboards. Split views help compare local geometry; overlays reveal route movement, length changes, additions, removals, and topology differences without losing the surrounding harness.
Visual highlighting should distinguish match confidence from change severity. A perfectly matched connector with a critical cavity reassignment is different from an uncertain entity match with no confirmed field delta.
The review must retain baseline value, current value, units, numeric delta where applicable, source revision, match evidence, acceptance state, and any linked finding or waiver.
A controlled comparison workflow
- Normalize both sources. Import each revision into the same neutral engineering model while retaining its provenance.
- Run readiness checks independently. Source-quality issues should not be confused with changes introduced by the revision.
- Generate and review entity matches. Confirm ambiguous pairs before interpreting additions, removals, and modifications.
- Inspect changes by domain. Filter electrical, physical, configuration, material, and documentary deltas according to reviewer responsibility.
- Trace impact. Follow changed entities through connected wires, segments, connectors, configurations, and applicable rules.
- Accept, reject, or defer explicitly. Record the disposition and reviewer evidence rather than treating “viewed” as approval.
- Export the review record. Deliver the field-level report and selected visual evidence with the released harness.
What belongs in the final report
- Unambiguous baseline and current revision identifiers.
- Match statistics, unresolved candidates, and matching assumptions.
- Added, removed, and modified entities grouped by engineering domain.
- Before and after values, units, numeric deltas, and physical locations.
- New, resolved, and unchanged validation findings and waivers.
- Reviewer dispositions, timestamps, comments, and export provenance.
Harness Studio compares neutral harness revisions using field-level deltas, synchronized split and overlay views, physical navigation, and exportable engineering evidence. Because both revisions can originate from different supported formats, the review focuses on the harness rather than the serialization.