Bug #37305 select count(*); returns value

Description:
count(*) returns value even if no tables are involved. I think this is wrong because:

mysql> select * from dual;
ERROR 1096 (HY000): No tables used
mysql> select *;
ERROR 1096 (HY000): No tables used
mysql> select count(*) from dual;
+----------+
| count(*) |
+----------+
|        1 |
+----------+
1 row in set (0.00 sec)

mysql> select count(*) from Q;
ERROR 1146 (42S02): Table 'test.Q' doesn't exist

mysql> select count(*) fromQ;
+-------+
| fromQ |
+-------+
|     1 |
+-------+
1 row in set (0.00 sec)

A single missing space would cause a bad query to actually work and return a bogus answer.

How to repeat:
mysql> select count(*) from dual;
+----------+
| count(*) |
+----------+
|        1 |
+----------+
1 row in set (0.00 sec)

mysql> select count(*);
+----------+
| count(*) |
+----------+
|        1 |
+----------+
1 row in set (0.00 sec)

Suggested fix:
Return "no tables used" if no tables are specified but the * argument to COUNT() is used.

Thank you for the bug report.

re:

"
mysql> select count(*) fromQ;
+-------+
| fromQ |
+-------+
|     1 |
+-------+
1 row in set (0.00 sec)

A single missing space would cause a bad query to actually work and
return a bogus answer.
"

The "bad query " is not necessarily a bad query. It's a query of the form "SELECT value alias_name", equivalent to "SELECT value AS alias_name".

Yes I understand the parsing problem with the AS being optional. I just wanted to point out that there are cases where allowing a standalone count(*) to return 1 can really bite the user.

This bug is verified, it was mistakenly left in "Analysing".

It might be debatable whether COUNT(*) without a table should return 0 or 1,
but as PostgreSQL and SQLite behave the same way I'd say: not a bug

Bug #87051 marked as duplicate of this one