Kari Lavikka <[EMAIL PROTECTED]> writes:
> --
> -- Choose a random point between 0 and max_uid and select the first
> -- value from the bigger part
> --
> CREATE OR REPLACE FUNCTION random_uid() RETURNS int4 AS
> 'SELECT uid FROM users u WHERE u.status = ''a'' AND uid >=
> cast(cast(max_uid() - 1 AS FLOAT) * random() AS INTEGER) ORDER BY uid
> ASC LIMIT 1'
> LANGUAGE 'sql';
This isn't going to do what you think because the random() function is
re-evaluated at every row of the table. (For that matter, so is
max_uid(), which means performance would suck even if it worked ...)
I'd suggest rewriting in plpgsql so you can assign the (max_uid-1)*random()
expression to a variable and then just use the variable in the SELECT.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 7: don't forget to increase your free space map settings
