database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-09-09 13:09:39 +03:00
Code Activity

Relax transactional restrictions on ALTER TYPE ... ADD VALUE (redux).

Browse Source 
Originally committed as 15bc038f (plus some follow-ups), this was
reverted in 28e07270 due to a problem discovered in parallel
workers.  This new version corrects that problem by sending the
list of uncommitted enum values to parallel workers.

Here follows the original commit message describing the change:

To prevent possibly breaking indexes on enum columns, we must keep
uncommitted enum values from getting stored in tables, unless we
can be sure that any such column is new in the current transaction.

Formerly, we enforced this by disallowing ALTER TYPE ... ADD VALUE
from being executed at all in a transaction block, unless the target
enum type had been created in the current transaction.  This patch
removes that restriction, and instead insists that an uncommitted enum
value can't be referenced unless it belongs to an enum type created
in the same transaction as the value.  Per discussion, this should be
a bit less onerous.  It does require each function that could possibly
return a new enum value to SQL operations to check this restriction,
but there aren't so many of those that this seems unmaintainable.

Author: Andrew Dunstan and Tom Lane, with parallel query fix by Thomas Munro
Reviewed-by: Tom Lane
Discussion: https://postgr.es/m/CAEepm%3D0Ei7g6PaNTbcmAh9tCRahQrk%3Dr5ZWLD-jr7hXweYX3yg%40mail.gmail.com
Discussion: https://postgr.es/m/4075.1459088427%40sss.pgh.pa.us
This commit is contained in:
Thomas Munro
2018-10-09 12:51:01 +13:00
parent 7767aadd94
commit 212fab9926
12 changed files with 367 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
src/test/regress/sql/enum.sql
View File

@@ -273,19 +273,34 @@ ALTER TYPE rainbow RENAME VALUE 'blue' TO 'green';
--
CREATE TYPE bogus AS ENUM('good');
-- check that we can't add new values to existing enums in a transaction
-- check that we can add new values to existing enums in a transaction
-- but we can't use them
BEGIN;
ALTER TYPE bogus ADD VALUE 'bad';
ALTER TYPE bogus ADD VALUE 'new';
SAVEPOINT x;
SELECT 'new'::bogus; -- unsafe
ROLLBACK TO x;
SELECT enum_first(null::bogus); -- safe
SELECT enum_last(null::bogus); -- unsafe
ROLLBACK TO x;
SELECT enum_range(null::bogus); -- unsafe
ROLLBACK TO x;
COMMIT;
SELECT 'new'::bogus; -- now safe
SELECT enumlabel, enumsortorder
FROM pg_enum
WHERE enumtypid = 'bogus'::regtype
ORDER BY 2;
-- check that we recognize the case where the enum already existed but was
-- modified in the current txn
-- modified in the current txn; this should not be considered safe
BEGIN;
ALTER TYPE bogus RENAME TO bogon;
ALTER TYPE bogon ADD VALUE 'bad';
SELECT 'bad'::bogon;
ROLLBACK;
-- but ALTER TYPE RENAME VALUE is safe in a transaction
-- but a renamed value is safe to use later in same transaction
BEGIN;
ALTER TYPE bogus RENAME VALUE 'good' to 'bad';
SELECT 'bad'::bogus;
@@ -293,12 +308,21 @@ ROLLBACK;
DROP TYPE bogus;
-- check that we *can* add new values to existing enums in a transaction,
-- if the type is new as well
-- check that values created during CREATE TYPE can be used in any case
BEGIN;
CREATE TYPE bogus AS ENUM();
ALTER TYPE bogus ADD VALUE 'good';
ALTER TYPE bogus ADD VALUE 'ugly';
CREATE TYPE bogus AS ENUM('good','bad','ugly');
ALTER TYPE bogus RENAME TO bogon;
select enum_range(null::bogon);
ROLLBACK;
-- ideally, we'd allow this usage; but it requires keeping track of whether
-- the enum type was created in the current transaction, which is expensive
BEGIN;
CREATE TYPE bogus AS ENUM('good');
ALTER TYPE bogus RENAME TO bogon;
ALTER TYPE bogon ADD VALUE 'bad';
ALTER TYPE bogon ADD VALUE 'ugly';
select enum_range(null::bogon); -- fails
ROLLBACK;
--