cadrumo.domain.calculations.registry.cleared_families module¶
An edition’s authored statement that it takes none of a family from its predecessor.
A delta edition inherits what it does not restate, family by family. Two declarations decline that, and they decline different amounts. A restatement says “I state this family in full here”; the members are the successor’s own and the predecessor’s are simply not consulted. A clearance says the stronger thing: this edition has NONE of this family, and the predecessor’s members do not carry into it.
The clearance is the stronger claim and so needs at least as much behind it. It names the family, the cause, and the reason somebody wrote. Absence cannot carry it: an edition that inherits and an edition that meant to take nothing look identical from the successor’s files alone, which is precisely why the loader refuses an undecided scoped family rather than resolving it.
The cause matters more here than it does for a restatement, because two very different situations empty a family and only one of them is a statement about the law. Recording which one keeps an unfinished surface from reading as a legal withdrawal.
The family is closed to CLEARABLE_FAMILIES: a clearance is applied by
the keyed merge, so naming a family that merge never carries declines nothing
while reading as though it did.
- CLEARABLE_FAMILIES: Final[frozenset[str]]¶
Every family the keyed merge carries by member identity, and so the only families a clearance can empty.
casillasinherit through the loader’s own casilla pass, which does not read this declaration; a clearance naming them would validate while the merge still carried every row. The set is projected from the domain-owned policy table so the authored claim and the merge stay closed on one vocabulary.
- cleared_family_names(declarations)[source]¶
Return the families a
cleared_familiesvalue withdraws.Compilation, the delta signal and the migration assessment all need the same answer from the same value, and they hold it at different stages: the loader sees raw TOML tables, later consumers may hold typed declarations. Reading both shapes here keeps one answer rather than a membership test per consumer, which is how a bare-string reading survived in three places at once.
Nothing is validated. An entry that is neither shape, or one naming no family, contributes nothing and is refused by
ClearedFamilyDeclarationitself, so a reading that recognises nothing declines nothing and lets the edition be refused with the schema’s error rather than a consumer’s.
- class ClearedFamilyCause(*values)[source]¶
Bases:
StrEnumWhy an edition takes none of a family from its predecessor.
The two members are not degrees of the same thing. One is a claim about the governing document, the other is a claim about this repository’s authoring state, and collapsing them would let unfinished work read as law.
- OFFICIAL_STRUCTURE_WITHDRAWS¶
the predecessor’s members are withdrawn BY that document, so carrying them would assert a structure the edition’s own authority contradicts.
- Type:
The document governing this edition no longer lays the family out
- NOT_AUTHORED_FOR_THIS_EDITION¶
Nobody has authored this edition’s own members yet, and the predecessor’s are not adopted in their place. This is an acknowledged capability gap, not a legal statement: it says the registry has nothing here, and says it where a reader and a coverage report can both see it. A family genuinely outside the modelo’s legal requirements is a family disposition instead, which is a claim about the law and is validated against the predecessor chain as one.