Oracle SQL

Work with unbounded lists in PL/SQL with nested tables, handle gaps after deletions, and pass lists of values to SQL with TABLE().

Store values under string or integer keys with associative arrays, use them as fast in-memory lookups, and handle missing keys safely.

Return one or more result sets from a PL/SQL block or procedure to the client with DBMS_SQL.RETURN_RESULT, without OUT cursor parameters.

Reuse one explicit cursor for many similar queries by giving it parameters with defaults, passed by position or by name.

Find out how many rows an UPDATE or DELETE changed, or whether it changed any, with the implicit cursor attributes of PL/SQL.

Step by any amount, cover several ranges, filter iterations, and loop over collection values, indexes, or pairs in one PL/SQL FOR loop.

Give PL/SQL types meaningful names with built-in limits, such as a 0 to 100 percentage, and let the language enforce them on every assignment.

Mark points in a transaction with SAVEPOINT and undo only the changes made after one with ROLLBACK TO SAVEPOINT, keeping earlier work.

Get the values before and after an update in the same statement with RETURNING OLD and NEW, for a single row or many with BULK COLLECT.

Keep a bulk INSERT or UPDATE going when some rows fail: log the bad rows to an error table with LOG ERRORS and load the rest.

Speed up bulk loads and archiving with direct-path INSERT and the APPEND hint, and confirm the load path with LOAD AS SELECT in the plan.

Route the rows of one query into several tables in a single statement with INSERT ALL and INSERT FIRST, with conditions or without.

See the live state of the database with V$ views: find your session, read parameter values, check the version, and list the PDBs.

Find every constraint of a table, what each foreign key references, and which tables point to a table, with USER_CONSTRAINTS.

Answer questions about your schema in SQL with the data dictionary: find the right view, list objects and columns, and read comments.

Give developers exactly the privileges to build application objects with DB_DEVELOPER_ROLE, instead of CONNECT, RESOURCE, or DBA.

Give users access to every table of one schema, including future ones, with schema privileges instead of per-table or database-wide grants.

Give application users exactly the access they need with GRANT, down to single columns, and take it back with REVOKE when it is no longer needed.