Hi Daniel,
On Fri, Mar 14, 2025 at 2:26 PM Daniel Gustafsson <daniel@yesql.se> wrote:
>
> On 13 Mar 2025, at 15:03, Greg Sabino Mullane <htamfids@gmail.com> wrote:
>
> Patch looks good. One minor issue:
>
> greg=# \set WATCH_INTERVAL -1
> invalid value "-1" for "WATCH_INTERVAL": must be greater than 0.00
> greg=# \set WATCH_INTERVAL 0.00
> greg=#
>
> We should disallow 0 as the error message implies
>
>
> Ah, nice catch, fixed in the attached along with a test for the minimum bound (ie zero).
#\watch c=4 i=
Mon 17 Mar 2025 05:52:50 PM IST (every 0s)
?column?
----------
1
(1 row)
Mon 17 Mar 2025 05:52:50 PM IST (every 0s)
?column?
----------
1
(1 row)
Mon 17 Mar 2025 05:52:50 PM IST (every 0s)
?column?
----------
1
(1 row)
Mon 17 Mar 2025 05:52:50 PM IST (every 0s)
?column?
----------
1
(1 row)
0 is an accepted value for interval, even though it might look insensible.
The behaviour should be same in both cases \watch i=<some value> and
\set WATCH_INTERVAL <some value> \watch. In this case it's not. In
fact, we should use the same validation code in both the cases. Why
don't we perform the same additional validations in
ParseVariableDouble() in exec_watch_command() as well?
The test only validate default variable value. We need a test where we
see variable value being honored lie tests between 369 to 376 in the
same file.
--
Best Wishes,
Ashutosh Bapat
