Planned features
These are plans, not current features.
Source formats
- PostgreSQL
pg_dump. Plain SQL dumps, withCOPY … FROM stdindata and column types taken fromCREATE TABLE. One file then captures a whole database. - MySQL
mysqldumpINSERTstatements. - JSON Lines, for NoSQL and API dumps. This first needs a raw model for nested records.
- Avro and ORC.
- Direct JDBC/ODBC table dumps. Full table scans only, no views.
- Partitioned datasets. A directory of Parquet or CSV parts captured as one view.
Modelling
- Nested values (lists, structs, maps) as raw structure instead of text.
- Formula translation. Compile column formulas into SHACL rules and constraints.
- Typed value copies. Optionally, a typed copy of each value that conforms to its column's inferred datatype, next to the source literal.