Dong Wook Lee <sh95119@gmail.com> writes:
> On 22/06/18 12:07오후, Michael Paquier wrote:
>> As a whole, I am not sure that this is really worth spending cycles on
>> when running check-world or similar, and the code of the extension is
>> trivial.
> Even though it is trivial, I think it would be better if there was a test.
I looked at this and concur with Michael's evaluation. A new TAP module
is quite an expensive thing, since it incurs (at least) an initdb run.
In this case, the need to delay a long time to ensure that the test
doesn't fail on slow systems makes that even worse. I don't think
I want to incur these costs every time I run check-world in order to
test a pg_usleep() call, which is what this module boils down to.
If we had some sort of "attic" of tests that aren't run by either
check-world or most buildfarm members, perhaps this would be worth
putting there. But we don't.
One idea could be to install the test but leave the TAP_TESTS line in
the Makefile commented out. Then, somebody who was actively working on
the module could enable the test easily enough (without even modifying
that file: just do "make check TAP_TESTS=1"), but otherwise we don't
pay for it. However, I'm not sure how well that plan will translate
to the upcoming meson build system.
If we don't do it like that, I'd vote for rejecting the patch.
regards, tom lane