Cursors

On this page Carat arrow pointing down

A cursor is a placeholder into a selection query that allows you to iterate over subsets of the rows returned by that query.

Tip:

This document describes cursor usage within SQL transactions. For information about using cursors in PL/pgSQL functions and procedures, see PL/pgSQL.

Cursors differ from keyset pagination and LIMIT/OFFSET in that:

  • Each cursor is a stateful SQL object that is referred to by a unique name.
  • Each cursor requires holding open its own dedicated (read-only) transaction.
  • Each cursor operates on a snapshot of the database at the moment that cursor is opened.

Synopsis

Cursors are declared and used with the following keywords:

CLOSE ALL cursor_name

Examples

These examples assume the presence of the MovR data set.

Use a cursor

icon/buttons/copy
BEGIN;
DECLARE rides_cursor CURSOR FOR SELECT * FROM movr.rides;
icon/buttons/copy
FETCH FORWARD 5 FROM rides_cursor;
                   id                  |     city      | vehicle_city  |               rider_id               |              vehicle_id              |        start_address        |        end_address         |     start_time      |      end_time       | revenue
---------------------------------------+---------------+---------------+--------------------------------------+--------------------------------------+-----------------------------+----------------------------+---------------------+---------------------+----------
  8ccccccc-cccc-4000-8000-000000000113 | san francisco | san francisco | 80000000-0000-4000-8000-000000000019 | 77777777-7777-4800-8000-000000000007 | 69313 Jody Tunnel Apt. 17   | 5210 Kim Canyon Suite 84   | 2018-12-22 03:04:05 | 2018-12-22 07:04:05 |   36.00
  8d4fdf3b-645a-4000-8000-000000000114 | san francisco | san francisco | 80000000-0000-4000-8000-000000000019 | 88888888-8888-4800-8000-000000000008 | 54797 Lauren Cliffs Apt. 37 | 7425 Matthews Harbors      | 2018-12-18 03:04:05 | 2018-12-20 04:04:05 |   45.00
  8dd2f1a9-fbe7-4000-8000-000000000115 | san francisco | san francisco | 75c28f5c-28f5-4400-8000-000000000017 | 77777777-7777-4800-8000-000000000007 | 23053 Brown Creek           | 15838 Preston Unions       | 2018-12-26 03:04:05 | 2018-12-27 15:04:05 |   34.00
  55810624-dd2f-4c00-8000-0000000000a7 | seattle       | seattle       | 570a3d70-a3d7-4c00-8000-000000000011 | 55555555-5555-4400-8000-000000000005 | 78340 Ashley Common Apt. 4  | 19798 Riggs Spring         | 2018-12-08 03:04:05 | 2018-12-10 06:04:05 |   13.00
  56041893-74bc-4c00-8000-0000000000a8 | seattle       | seattle       | 570a3d70-a3d7-4c00-8000-000000000011 | 66666666-6666-4800-8000-000000000006 | 6431 Robert Forest          | 83655 Michael Cape Apt. 94 | 2018-12-09 03:04:05 | 2018-12-09 14:04:05 |   48.00
(5 rows)
icon/buttons/copy
CLOSE rides_cursor;
icon/buttons/copy
COMMIT;

Use a holdable cursor

By default, a cursor closes when the transaction ends. The WITH HOLD clause defines a holdable cursor, which stays open after a COMMIT by writing its results into a buffer. Use WITH HOLD to access data across multiple transactions without redefining the cursor.

Note:

The WITHOUT HOLD clause specifies the default non-holdable cursor behavior.

A holdable cursor can be opened in both explicit and implicit transactions. The following example uses a holdable cursor to return vehicles that are available for rides.

Start a transaction:

icon/buttons/copy
BEGIN;

Declare a cursor using WITH HOLD to keep it open after the COMMIT:

icon/buttons/copy
DECLARE available_vehicles_cursor CURSOR WITH HOLD FOR
  SELECT id, type, city, status FROM vehicles WHERE status = 'available';

Fetch the first two rows from the cursor:

icon/buttons/copy
FETCH 2 FROM available_vehicles_cursor;
                   id                  |  type   |   city    |  status
---------------------------------------+---------+-----------+------------
  bbbbbbbb-bbbb-4800-8000-00000000000b | scooter | amsterdam | available
  22222222-2222-4200-8000-000000000002 | scooter | boston    | available

Commit the transaction:

icon/buttons/copy
COMMIT;

Continue fetching rows from the cursor:

icon/buttons/copy
FETCH 2 FROM available_vehicles_cursor;
                   id                  |    type    |  city   |  status
---------------------------------------+------------+---------+------------
  33333333-3333-4400-8000-000000000003 | bike       | boston  | available
  55555555-5555-4400-8000-000000000005 | skateboard | seattle | available

Close the cursor:

icon/buttons/copy
CLOSE available_vehicles_cursor;

View all open cursors

icon/buttons/copy
SELECT * FROM pg_cursors;
            name            |                               statement                                | is_holdable | is_binary | is_scrollable |         creation_time
----------------------------+------------------------------------------------------------------------+-------------+-----------+---------------+--------------------------------
  rides_cursor              | SELECT * FROM movr.rides                                               |      f      |     f     |       f       | 2025-05-07 21:12:53.32978+00
  available_vehicles_cursor | SELECT id, type, city, status FROM vehicles WHERE status = 'available' |      t      |     f     |       f       | 2025-05-07 21:12:59.605647+00
(2 rows)

Known limitations

CockroachDB implements SQL cursor support with the following limitations:

  • DECLARE only supports forward cursors. Reverse cursors created with DECLARE SCROLL are not supported. #77102
  • FETCH supports forward, relative, and absolute variants, but only for forward cursors. #77102
  • BINARY CURSOR, which returns data in the Postgres binary format, is not supported. #77099
  • Scrollable cursor (also known as reverse FETCH) is not supported. #77102
  • SELECT ... FOR UPDATE with a cursor is not supported. #77103
  • Respect for SAVEPOINTs is not supported. Cursor definitions do not disappear properly if rolled back to a SAVEPOINT from before they were created. #77104

Differences between cursors and keyset pagination

Cursors are stateful objects that use more database resources than keyset pagination, since each cursor holds open a transaction. However, they are easier to use, and make it easier to get consistent results without having to write complex queries from your application logic. They do not require that the results be returned in a particular order (that is, you don't have to include an ORDER BY clause), which makes them more flexible.

Keyset pagination queries are usually much faster than cursors since they order by indexed columns. However, in order to get that performance they require that you return results in some defined order that can be calculated by your application's queries. Because that ordering involves calculating the start/end point of pages of results based on an indexed key, they require more care to write correctly.

See also


Yes No
On this page

Yes No