Temporal compile-time wiring:
- Rejects accepting the period attribute as action input — a temporal write always
establishes validity from
as_ofonward, so the period is set by the data layer, never by the caller. - Bakes a
range_overlaps(parent(source), destination)filter into relationships that declaretemporal_keyswith both a source and destination period attribute.
The relationship filter is done at transform time (not by checking the destination's
temporality, which would introduce a compile-time dependency) — it reads only the
declared temporal_keys option and attribute names. Because the overlap becomes part
of relationship.filter, every consumer (in-memory loads and data-layer joins)
applies it automatically. A separate verifier checks temporal_keys is correct.
Summary
Functions
Callback implementation for Spark.Dsl.Transformer.after?/1.
Callback implementation for Spark.Dsl.Transformer.after_compile?/0.
Callback implementation for Spark.Dsl.Transformer.before?/1.
Callback implementation for Spark.Dsl.Transformer.transform/1.
Functions
Callback implementation for Spark.Dsl.Transformer.after?/1.
Callback implementation for Spark.Dsl.Transformer.after_compile?/0.
Callback implementation for Spark.Dsl.Transformer.before?/1.
Callback implementation for Spark.Dsl.Transformer.transform/1.