LLKV Plan

Work in Progress

llkv-plan defines the typed plan surface that bridges parsed SQL and LLKV runtime execution. It organizes statement plans, expression programs, and subquery metadata so lower layers can execute without re-parsing SQL.

Responsibilities

Translate sqlparser ASTs into strongly typed plan enums for DDL, DML, and queries.
Encode correlated subqueries, scalar programs, and parameter bindings used during execution.
Provide canonical representations (e.g., CanonicalRow, CanonicalScalar) for hashing and deduplicating expression inputs.

Select Plan Structure

SelectPlan carries projection lists, filters, sort clauses, limits, and expression programs compiled from llkv-expr.
Correlated subqueries are captured as ScalarSubquery or FilterSubquery entries, pairing placeholder metadata with outer column bindings.
Join plans describe join type, keys, and required columns so llkv-executor can assemble row streams efficiently.

DML and DDL Plans

InsertPlan encapsulates Arrow RecordBatch sources, conflict policies, and column mappings consumed by llkv-runtime.
Update/Delete plans carry predicate programs plus column projections for the runtime to apply changes with MVCC bookkeeping.
Catalog operations (create table, create index, etc.) surface schema definitions used by llkv-table and the system catalog.

Correlation Utilities

llkv_plan::correlation centralizes placeholder generation and column mapping for correlated subqueries, ensuring planner decisions remain consistent across statements.
Shared helpers prevent duplication and keep higher layers focused on execution logic rather than subquery plumbing.

License