Besides several bug fixes, this is the first version with full sqlalchemy test coverage/support for EXASOL 5.0.
This release should speedup database reflection a lot. Instead of querying table and constraint information per table/row it caches the results per schema.
Reflecting an example schema with 12 tables went down from 14 seconds and 84 database queries to 1,4 seconds and 5 database queries.
The caching only works if you use the same inspector for all reflections:
from sqlalchemy import * from sqlalchemy.engine import reflection dsn = "exa+pyodbc://XXX" engine = create_engine(dsn) engine.echo = True insp = reflection.Inspector.from_engine(engine) md = MetaData(bind=engine) schemaname = "testschema" for tablename in insp.get_table_names(schema=schemaname, order_by="foreign_key"): tb = Table(tablename, md, schema=schemaname) insp.reflecttable(tb, None)
metadata.reflect(..) does not benefit that much from the caching, because sqlalchemy internally creates a new inspector instance for each table.