WireSpot wrote:
> I'm trying to use prepared statements in an application and I'm
> running into this error: "Query failed: prepared statement already
> exists".
>
> The reason is obvious. What I want to know is the best way to avoid
> getting this error. The client application sets statement names as MD5
> of the actual query text, which means there's a possibility of a clash
> between different parts of the applications if they attempt to prepare
> the same query in the lifetime of a connection.
>
> Possible solutions, please advise:
>
> 1) Something like PREPARE IF NOT EXISTS. I'm guessing no such luck.
>
> 2) Tweaking the Postgres error reporting to ignore this particular
> error. Is it possible? From a non-priviledged client connection?
>
> 3) Reading a list of all the currently defined prepared statements to
> see if the one I want is already prepared. I'm hoping some "magic"
> SELECT in pg's internal tables may do the trick. But I also worry
> about introducing overhead this way.
>
> I also imagined some workarounds in the code (PHP), such as defining a
> global/static hash table and registering statement names with it. But
> I'd like to know if there's a better way.
Do you still need the old prepared statement?
If not, you can simple DEALLOCATE it and then try the PREPARE again.
Something like that
try {
PREPARE statementnam AS SELECT ....;
} catch (SQLException e) {
if (e.getSQLState().equals("42P05")) {
DEALLOCATE statementnam;
PREPARE statementnam AS SELECT ....;
} else
throw e;
}
(that's Java pseudocode, but I hope you'll understand what I mean).
If you still need the old statement, generate a new, different name
and try again.
Yours,
Laurenz Albe