Appearance
ADR-0024: An SDS with no revision date is admitted, and the absence is shown
Status: Accepted — amends ADR-0013 Date: 2026-09-02
Decisions
D-51 Revision date is an optional part of Chemical Record identity, and its absence is displayed
Identity is product name + supplier, plus revision date when the document states one. A record without one displays "revision date not stated", and D-06's staleness indicator shows unknown rather than a substituted value.
Context
R-41: one of three spike documents carries no revision date anywhere in eight pages, so D-18's product + supplier + revision-date triple cannot be formed and the staleness signal has nothing to read.
This is the D-03 pattern applied to metadata: "Not Stated" is a correct answer. R-43 found that real SDS text says as much about itself in places, so the product is already displaying supplier-authored absence elsewhere.
Rejected options
- Substitute the Curation date or the file timestamp — every record would gain a usable staleness signal, and it fabricates provenance while displaying it with the confidence of a real one. Same class of error as merging records (D-45), and worse because it looks authoritative.
- Reject undated SDS from the Corpus — keeps identity clean, rejected because The Plant physically holds that drum: an app that cannot answer for a chemical standing in the store because its paperwork is untidy fails at the moment it matters.
- Admit but mark the whole record degraded — rejected as deterring use of a record that is otherwise complete.
This ruling may not be re-decided
If a change contradicts this ADR: stop and raise it. Do not implement over it. Never populate a revision date the document does not state.
Consequences
Two revisions of an undated product collide on the same identity. The curator must notice; the product cannot.
Coverage
| Upstream | Landed in | Evidence | Note |
|---|---|---|---|
| R-41 | D-51 | the undated document is named in Context | |
| R-43 | D-51 | supplier-authored absence already appears in real text |