Heikki Linnakangas wrote:
Pedro Gimeno wrote:
> When a function has a SQL statement to execute that has an
> unqualified table, that SQL statement doesn't always pick the table
> from a schema in the search_path.
The first time the function is run, all the statements in it are
planned and the schema used for the insert is resolved. Subsequent
calls will reuse the same plan. The obvious fix is to always
explicitly specify the schema in the INSERT. Or if if you want it to
depend on the search_path, you can execute it dynamically using
EXECUTE
(http://www.postgresql.org/docs/8.2/static/plpgsql-statements.html#PLPGSQL-STATEMENTS-EXECUTING-DYN).
That's the workaround I'm using, yet I find this behaviour quite
unexpected. Using EXECUTE prevents the benefit of using prepared
statements until search_path changes or a schema is deleted or renamed,
not to mention the impact on readability.
What I expect is that when search_path changes, either explicitly or
implicitly (e.g. by creating a temporary table), or a schema is deleted
or renamed, the affected prepared queries are parsed again so that the
right schema is picked.
This would allow using a common schema for functions instead of
defining them once in every schema in which to apply them, in cases
(like mine) where the same operations have to be applied to different
schemas.
-- Pedro Gimeno
---------------------------(end of broadcast)---------------------------
TIP 7: You can help support the PostgreSQL project by donating at
http://www.postgresql.org/about/donate
