`CoherentSample.energyRegisterWh` (in `CoherentSampleComputer.ts`) is
computed from `connectorStatus.transactionEnergyActiveImportRegisterValue
+ deltaEnergyWh` (transaction-scoped) at sample-compute time, but the
emitted `Energy.Active.Import.Register` measurand reads
`connectorStatus.energyActiveImportRegisterValue` (station-scoped) via
`resolvePhasedValue` in `CoherentMeterValueBuilder.ts`. The two counters
diverge whenever the station has non-zero pre-transaction register
history, converging only when the transaction begins on a station whose
register was previously zero.
Documents the divergence on the field's JSDoc so tests and in-process
introspection paths reading `sample.energyRegisterWh` understand the
scope distinction. No runtime change.
* fix(meter-values): tighten L-L voltage guard to 3-phase-only (issue #1936 h.1)
`resolvePhasedValue`'s Line-to-Line voltage branch (in
`CoherentMeterValueBuilder.ts`) previously used `numberOfPhases <= 1` as
the skip guard, which meant that `numberOfPhases === 2` silently emitted
`Math.sqrt(2) * voltageV` -- a physically meaningless value.
2-phase AC is unsupported by contract across the codebase:
`Helpers.getPhaseRotationValue` branches only on `{0, 1, 3}` (AC 1-phase,
AC 3-phase, and DC), and the station-template validator enforces
numberOfPhases in `{0, 1, 3}`. The `<= 1` guard was written assuming the
enum-open case would default to the 3-phase computation; the 2-phase gap
escaped review because no valid station configuration reaches it in
production.
Tightens the guard to `numberOfPhases !== 3` so any non-3-phase
configuration (0, 1, 2, or hypothetical future values) returns
`undefined`, letting the caller log-and-skip rather than emit
`sqrt(2) * V_LN` nonsense. Same behavior for 3-phase (still emits
`sqrt(3) * V_LN`) and for 1-phase (still skipped). Adds a WHY comment
documenting the `V_LL = sqrt(3) * V_LN` contract.
The h.1 guard tightened `numberOfPhases <= 1` to `numberOfPhases !== 3`,
widening the L-L voltage skip predicate from "1-phase only" to "any
non-3-phase count". The `resolvePhasedValue` JSDoc header still
described the pre-h.1 skip case as "`sqrt(phases)` collapses to 1 on
single-phase, in which case L-L has no physical meaning and the
template is skipped" -- factually incomplete post-h.1 (N=0/2/4+ also
skip through the same predicate).
Updates the JSDoc to reflect the current predicate: "L-L is defined
only for balanced 3-phase AC; skipped for any other phase count".
Documentation-atomicity fix per AGENTS.md instruction to "update code,
tests, and documentation atomically".
No runtime change. Test invariant `fail: 0, skipped: 6` preserved
(2925 pass / 2931 total).
* style(meter-values): normalize prose em-dash and ellipsis to ASCII (issue #1936 h)
The session ASCII rule permits only the section sign (U+00A7) plus
pre-existing physics/math notation in JSDoc as non-ASCII exceptions.
It does not permit em-dash (U+2014) or horizontal ellipsis (U+2026)
as prose punctuation. Both symbols pre-existed in the 2 files touched
by PR-H (5 em-dashes in `CoherentMeterValueBuilder.ts`, 14 em-dashes
and 1 ellipsis in `CoherentSampleComputer.ts`) and were classified
through R1-R8 as SUB-THRESHOLD per scope-discipline convention
(pre-existing outside the line-level PR-H delta).
Under explicit user directive to normalize (Option B follow-up),
replaces all 20 prose Unicode occurrences with ASCII (U+2014 to `-`,
U+2026 to `...`) in the 2 touched files. Physics/math notation
(sqrt, multiplication dot, implication arrow, sigma, phi, element-of,
delta, less-or-equal, epsilon) preserved per pre-existing JSDoc
convention.
repositories / e-mobility-charging-stations-simulator.git / commit
