Michael Paquier <michael@paquier.xyz> writes:
> On Fri, Apr 09, 2021 at 08:52:20AM +0200, Fabien COELHO wrote:
>> There is not a single test of "ctrl-c" which would have caught this trivial
>> and irritating regression. ISTM that a TAP test is doable. Should one be
>> added?
> If you can design something reliable, I would welcome that.
+1, there's a lot of moving parts there.
I think avoiding any timing issues wouldn't be hard; the
query-to-be-interrupted could be "select pg_sleep(1000)" or so.
What's less clear is whether we can trigger the control-C
response reliably across platforms.
regards, tom lane