Re: Can not cancel a call to a function that has opened a refcursor

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Mike Knowsley <mike(dot)knowsley(at)bidorbuy(dot)co(dot)za>
Cc: pgsql-bugs(at)lists(dot)postgresql(dot)org
Subject: Re: Can not cancel a call to a function that has opened a refcursor
Date: 2021-08-05 15:19:09
Message-ID: 2392578.1628176749@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Mike Knowsley <mike(dot)knowsley(at)bidorbuy(dot)co(dot)za> writes:
> A request to cancel a call to a function does not cancel the function if the function has opened a refcursor which it is returning.

As far as I can see, there is no such bug; you're just misunderstanding
when the query referenced by the cursor will execute. The function
test_delay_in_refcursor() sets up the cursor and returns basically
immediately (certainly in much less than 2 seconds), so the timeout
you have in the JDBC code will never fire. The expected pg_sleep(5)
won't execute till you try to fetch something from the refcursor.

Trying this manually in psql, I get

...
CREATE FUNCTION
regression=# \timing
Timing is on.
regression=# begin;
BEGIN
Time: 0.306 ms
regression=*# select test_delay_in_refcursor();
test_delay_in_refcursor
-------------------------
<unnamed portal 1>
(1 row)

Time: 0.314 ms
regression=*# fetch from "<unnamed portal 1>";
?column? | pg_sleep
----------+----------
test |
(1 row)

Time: 5005.475 ms (00:05.005)

I could have canceled the FETCH (and doing so works);
but I'm way too old and slow to cancel the submillisecond
setup step.

regards, tom lane

In response to

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message David G. Johnston 2021-08-05 15:34:23 Re: BUG #17134: pg_restore ERROR: operator does not exist: util.ltree = util.ltree
Previous Message David G. Johnston 2021-08-05 14:35:11 Re: BUG #17135: duplicate key value violates unique constraint