Expand description
§SuperTable Partitioning
This module defines the partitioning strategy for SuperTable, supporting Hidden Partitioning.
§Hidden Partitioning
Unlike Hive-style partitioning where users must explicitly query partition columns
(e.g., WHERE year = 2024), SuperTable tracks the relationship between
data columns and partition values. Users query the data column (e.g., WHERE timestamp > ...)
and the query engine automatically prunes partitions based on the transform.
§Transforms
Supported transforms:
- Identity: Value is used as-is (e.g.,
category) - Bucket(N): Hash of value % N (e.g.,
bucket(user_id, 16)) - Truncate(W): Truncate string/binary to width W (e.g.,
truncate(name, 1)) - Year/Month/Day/Hour: Extract date/time component
Structs§
- Partition
Field - A field in a partition spec.
- Partition
Spec - A partition spec defines how a table is partitioned.
- Partition
Spec Builder
Enums§
- Transform
- Supported partition transforms.