Renames, paths and file identity
Distinguish unchanged content at a new path from content replacement at the old path.
Research note · November 14, 2026Research question
Should a renamed model file appear as unchanged, moved, or missing and added?
Method
- Create a fixture with two files sharing the same bytes at different paths. Rename one, then record the path-based and digest-based interpretations independently.
- Define whether the manifest represents a directory layout or an unordered set of content objects. A path-sensitive loader can behave differently after a rename even if every digest still exists somewhere.
- Test case changes, Unicode filenames and path separators on the supported platforms. Preserve original paths and report ambiguous mappings instead of selecting a convenient match.
Records to retain
- Original and revised path maps
- Content digest groups
- Platform and filesystem behavior
- Ambiguous rename candidates
Source basis
A content digest identifies bytes, while a provenance record can identify an entity within a larger process. Neither determines whether two path occurrences are interchangeable for a specific loader. Rename detection should be labeled as interpretation.
Interpretation & limits
Equal content is not proof that a move was intended. Filesystem normalization differs across systems, and this note does not prescribe a universal path-normalization rule.
Next experiments
- Test a case-sensitive fixture on every supported platform.
- Measure how reviewers distinguish a move from a replacement.
References
Technical reference
Implementation details supporting this note, drawn directly from the maintained documentation. These are technical references, not additional papers or new experimental results. Expand a topic to inspect its method, fields and limits.