Your original question says that the function returns the number of rows deleted. That's one line. Now you're saying that it returns one row per table-row deleted. Which is it?
On 7/7/20 7:21 PM, Wells Oliver wrote:
ha, the CTE approach to only get one line of output versus however many hundreds of rows were used for the delete is perfect. Thanks.
Vanilla SQL script calls a plpgsql function to delete some number of rows from three tables:
SELECT mydelete(r) FROM sometable;
Where sometable contains maybe 100+ records. This causes the results from the function (integer of number of rows removed) to be displayed in the output, like you'd kinda expect with a SELECT call, except I don't want to see it all, I just want the function quietly executed and rows removed.
Can I accomplish this?
Pure SQL, no, you cannot just ignore the output. You can perform post-processing (via CTE/WITH) to reduce how much is printed (aggregates). If you are using psql you can send it to /dev/null. You could use a DO block and (kinda) ignore the result (SQL) and/or stick it into a throw-away variable (plpgsql).