OptionalgapWhy no value exists; required exactly when value is null.
Open metric key, such as workplane.total.illuminance.
OptionalstatusWhich kind of absence a null value is: unsupported when the solver cannot
compute it at all, not-run (the default) when this study lacked an input.
Targets the production declared, already checked for shape and uniqueness.
Unit the value is produced in, compared exactly with a target's unit.
Finite measured value, or null when the solver produced none.
Caller-owned sink the resolver appends a unit-mismatch warning to.
Build one metric, resolving whatever target the production declared for it.
A target stated in the wrong unit does not silently pass and does not silently fail: it is dropped and reported as a
target-unit-mismatchwarning, so the metric readsuntargeted(which is true, nothing comparable was declared) while the author is told exactly which declaration is wrong. Comparing 300 lux against a target of 300 candela would be the alternative, and it would clear a room nobody measured correctly.A metric with no value must say why. Passing a value and a gap together, or neither, throws: those are the two shapes that let an absent measurement masquerade as a result. A non-finite value also throws. A metric with no value never carries a target, whatever was declared for its key.
Units are compared as exact strings with no conversion, so
lxandluxare different units. The first target with the metric's key decides; duplicate keys are refused earlier byassertAutoMovieAnalysisTargets.