information_schema view follows Databricks’ INFORMATION_SCHEMA: its column names and order. Open-source Spark has no information_schema, so both modes use the Databricks shape.
Supported views
information_schema.catalogs(Databricks reference).information_schema.catalog_privileges(Databricks reference).information_schema.columns(Databricks reference).information_schema.information_schema_catalog_name(Databricks reference).information_schema.key_column_usage(Databricks reference). Always empty.information_schema.routines(Databricks reference). Always empty: built-in functions are not listed.information_schema.schema_privileges(Databricks reference).information_schema.schemata(Databricks reference).information_schema.table_constraints(Databricks reference). Always empty.information_schema.table_privileges(Databricks reference).information_schema.tables(Databricks reference).information_schema.views(Databricks reference).
INFORMATION_SCHEMA view not listed does not exist in these modes: querying it raises a “relation does not exist” error, even where Firebolt has a native view of that name.
Addressing
A Databricks catalog is a Firebolt database:INFORMATION_SCHEMA.TABLESandsystem.INFORMATION_SCHEMA.TABLES: the current database.<database>.INFORMATION_SCHEMA.TABLES: the named database.
Notes
- Values carry Firebolt conventions, not Databricks’:
columns.data_typeandcolumns.full_data_typereturn Firebolt type names (INTEGER,TEXT), andviews.view_definitionreturns Firebolt DDL. tablesandschematalist user objects only, never the system schemas.- Databricks columns with no Firebolt source are
NULL. - A view cannot be created over a Databricks-shaped
information_schemaview, because the stored definition could not be re-resolved outside the dialect. Query it directly, or useCREATE TABLE AS SELECT.