6 Completeness and honesty§
AGREED This clause is the reason the part exists. The formats are means.
6.1 The domain declaration§
The timeSeries domain of BUS-1 16.6 now declares against a specification. complete means every series the exporting system holds for the model’s points is present. partial means some are, and the reason code says why the rest are not. absent with a reason remains an honest package at any profile below full. deferredToLaterVersion is no longer a valid reason for this domain — the version arrived.
6.2 Reason codes for partial history§
Real historian exports fail partially, in recurring ways, and the reason codes are how each way is said in a fixed vocabulary an owner can read before purchase. Two codes are added to BUS-1 16.6 in this version because partial history needed statements the prior vocabulary could not make honestly:
| The situation | The declaration |
|---|---|
| The retention window purged it. 2015 was collected and is gone. | partial with retentionExpired. Distinct from notHeldBySystem because collected and purged and never collected age into different answers — the same distinction that makes provenance method REQUIRED (BUS-1 14.1). |
| The historian was unreachable at export time — offline, an archive tier unmounted. | partial or absent with sourceUnavailable. The data may exist; this package does not carry it, and neither of the older codes says that. |
| The owner excluded an interval — a tenancy dispute, a confidential period. | partial with excludedByOwner, and the interval named in a note. |
| The export would be too large and the history is available by another route. | partial or absent with excludedForSize (5.3). |
| The system trends the points; the export function cannot read the historian. | partial or absent with notSupportedByExporter — the code an owner should read as a product gap. |
| The system never trended these points. | notApplicable, or absent with notHeldBySystem. Nothing was lost; nothing existed. |
Test TS-7 checks a complete declaration against the package’s own point declarations, and EX-10 grades the rest against the source system during certification. Nothing in this clause can compel completeness. Like BUS-1 16.6, it converts an unbounded quality problem into a bounded disclosure problem and makes the disclosure uncomfortable to write.
6.3 Change-of-value data§
A change-of-value historian records instants of change, not a grid, and 3.5-1 requires it exported that way. The temptation this clause exists to name: resampling COV data onto fifteen-minute rows produces a file that looks like every other trend and is a fabrication — each invented row asserts an observation nobody made. A receiver interpolating COV data for display is doing analysis, which is its business; an exporter interpolating it into the package is manufacturing evidence.
6.4 Systems that hold only rollups§
Some systems discard raw records on a schedule and keep hourly summaries forever. The honest export is the rollup series with its aggregation declared (3.6), plus a domain declaration that says what happened to the raw records: retentionExpired where they were collected and purged, notHeldBySystem where summaries are all that was ever stored. What 3.5-1 and this clause together forbid is the third option — shipping the summaries undeclared, as if the building were observed once an hour.
6.5 Sources that stored local time§
Some historians store wall-clock local time with no offset. Requirement 4.4-1 still applies: the exporter converts to UTC using the IANA rules of the zone the source ran in, and records that zone in sourceTimeZone so the conversion is auditable and wall-clock time is reconstructable.
One hour a year cannot be converted with certainty: during a backward daylight-saving transition each local time occurs twice, and a bare local timestamp does not say which. An exporter MUST resolve the ambiguity by the earlier UTC instant, consistently, and SHOULD record the limitation in completeness.knownLimitations. This is a real loss, it is the source system’s loss rather than this format’s, and the honest response is to state the rule and the residue rather than pretend the conversion is exact.