Modules
A native Elixir DataFrame API for Apache Spark, over Spark Connect.
A CASE WHEN chain, waiting for otherwise/2.
Spark's catalog: databases, tables, views, caching.
Expressions: column references, literals, operators.
A lazy DataFrame: a session and an inert plan.
Anything Latu returns as {:error, _}.
What a run reported besides its result: the metrics you asked for, and the ones Spark keeps.
Spark's function library.
A DataFrame with grouping attached, waiting for agg/2.
A merge being assembled: a source frame, a target table, and the clauses to apply.
Relation and expression trees, built directly as protobuf. Pure: no session, no network, no IO.
What the server says it is doing, while it is still doing it.
Arrow batches in, Explorer out — and, for Latu.create_dataframe/3, the other direction.
When a failed RPC is tried again, and how long Latu waits in between.
Where the Spark Connect server is, and the identity it keys state on.
An expression that references other DataFrames, and the relations it references.
The :telemetry events Latu emits, and the only place their names are written down.
A window specification: how to partition, how to order, and which rows the frame covers.