SQLite

View Ticket
Login
Ticket Hash: 3b4450072511e6215466eba4237339ef70ca4f10
Title: Inconsistent column names
Status: Fixed Type: Code_Defect
Severity: Important Priority: Immediate
Subsystem: Unknown Resolution: Fixed
Last Modified: 2017-12-23 12:34:35
Version Found In: 3.21.0
User Comments:
drh added on 2017-12-22 19:28:56:

In the SQL below, the result-set column name for the first SELECT is "a" but the column name in the table created by the CREATE TABLE AS on an identical SELECT is "x".

CREATE TABLE t1(x INT);
INSERT INTO t1(x) VALUES(1);
.headers on
SELECT a FROM (SELECT x AS a FROM t1);
CREATE TABLE t2 AS
SELECT a FROM (SELECT x AS a FROM t1);
PRAGMA table_info(t2);

This problem first appeared with check-in [ade7ddf1998190b2] which is a fix for a prior ticket [de3403bf5ae5f72ed].


drh added on 2017-12-22 19:36:42:

Another example:

.headers on
CREATE TABLE t1(aaa INT);
INSERT INTO t1(aaa) VALUES(1);
SELECT BBb FROM (SELECT aaa AS Bbb FROM t1);
CREATE TABLE t2 AS
SELECT BBb FROM (SELECT aaa AS Bbb FROM t1);
PRAGMA table_info(t2);

For the straight SELECT statement, recent code gives a column name of Bbb and a column name of aaa for the t2 table. Older code gives the name BBb for both.