Re: [HACKERS] Re: [COMMITTERS] pgsql: Remove pgbench "progress" testpending solution of its timing is (fwd)

>>> Indeed… but then throttling would not be tested:-) The point of the test
>>> is to exercise all time-related options, including throttling with a
>>> reasonable small value.
>> 
>> Ok. I don't think that's really worthwhile. If we add some code that only 
>> runs in testing, then we're not really testing the real thing. I wouldn't 
>> trust the test to tell much. Let's just leave out that magic environment 
>> variable thing, and try to get the rest of the patch finished.
>
> If you remove the environment, then some checks need to be removed, because 
> the 2 second run may be randomly shorten when there is nothing to do. If not, 
> the test will fail underterminiscally, which is not acceptable. Hence the 
> hack. I agree that it is not beautiful.
>
> The more reasonable alternative could be to always last 2 seconds under -T 2, 
> even if the execution can be shorten because there is nothing to do at all, 
> i.e. remove the environment-based condition but keep the sleep.

Yet another option would be to replace the env variable by an option, eg
"--strict-time", that would be used probaly only by the TAP test, but 
would be an available feature.

-- 
Fabien.