Data — API Reference

Package: dodil.k3.tables.v1 · Service: TableService

The Data RPCs are typed shortcuts for row-level operations. Pass JSON rows / predicates / match-columns and K3’s planner routes the operation through the same strategies as a SQL Execute — they’re interchangeable.

When to use Data vs Execute: use Data when you’d rather hand the API typed inputs than assemble SQL strings (avoids quoting / templating / injection concerns). Use Execute for everything Data doesn’t cover — full SQL flexibility, joins, CTEs, window functions, CTAS, DDL.

RPC	HTTP	Routes through (planner strategy)	Page
`Query`	`POST /:bucket/tables/_query`	`QueryStrategy` — see Execute → Read strategies	Query
`Insert`	`POST /:bucket/tables/:table_name/insert`	`WRITE_STRATEGY_KEYED_INSERT_BULK` (or `NON_KEYED_INSERT` for tables without `merge_keys`)	Insert
`Merge`	`POST /:bucket/tables/:table_name/merge`	`WRITE_STRATEGY_MERGE_ROWS`	Merge
`Update`	`POST /:bucket/tables/:table_name/update`	`WRITE_STRATEGY_KEYED_UPDATE` (or `NON_KEYED_UPDATE` if the predicate doesn’t match `merge_keys` — ⚠️ WAL-bypass)	Update
`DeleteRows`	`POST /:bucket/tables/:table_name/delete-rows`	`WRITE_STRATEGY_KEYED_DELETE` (or `NON_KEYED_DELETE` — ⚠️)	DeleteRows

Sub-pages

Query — SELECT against a single target with optional broadcast JOINs (lookup_tables) and freshness control (EVENTUAL / STRONG / OLTP_ONLY).
Insert — append or overwrite JSON rows. Routes through the write log for keyed tables, direct-to-Delta for tables without merge_keys.
Merge — upsert via match-columns with when_matched / when_not_matched actions. rows_written counts source rows (pre-drain).
Update — column → value map + SQL WHERE predicate. Keyed predicates route through the write log; non-keyed predicates bypass the log (Delta-direct, WAL-bypass risk).
DeleteRows — predicate-driven delete. Keyed → tombstones in the write log; non-keyed → Delta-direct.

Data — API Reference

Sub-pages

See also