> On Nov. 22, 2013, 2:58 a.m., Vinod Kone wrote:
> > src/tests/master_contender_detector_tests.cpp, line 433
> > <https://reviews.apache.org/r/15710/diff/2/?file=389605#file389605line433>
> >
> > I don't know if I like this abstraction.
> >
> > How about using the AWAIT_ macro here? If that doesn't work with paused
> > clocks we should fix that instead.
>
> Ben Mahler wrote:
> I'm also in favor of this, we would have to enhance or add to the AWAIT_
> macros for paused clocks to loop advancing the clock (possibly by a provided
> duration) until the provided future transitions from pending.
> Such a macro would indeed be better than LOOP_FOR and would be very
> useful for tests that have a clock advance loop in paused tests.
>
> However, I'm interested in tests like isolator_test.cpp, which cannot use
> an AWAIT_ style macro. It currently does exactly what LOOP_FOR is doing. It's
> possible that this is the only example so perhaps we will want to avoid
> adding this for a single test.
>
> Jiang Yan Xu wrote:
> I think fixing AWAIT_ solves another type of problems but in this case
> (and in some in e.g. SlaveRecoveryTest) I can't use AWAIT_ either.
>
> You wait because a delay is involved and you want to advance the lock to
> avoid really wait for that long.
> You have to advance repeatedly (because you may have advanced the clock
> before delay is called).
>
> Otherwise you have to first wait for delay to get called which
> FUTURE_DISPATCH cannot accomplish.
Let's imagine you have something that handles a paused clock, like:
AWAIT_READY'(future, interval, limit)
{
Duration total;
if (Clock::paused()) {
while (future.isPending() && total < limit) {
Clock::advance(interval);
Clock::settle();
total += interval;
}
} else {
...
}
}
Would this not help in this test?
- Ben
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15710/#review29275
-----------------------------------------------------------
On Nov. 21, 2013, 11:18 p.m., Jiang Yan Xu wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15710/
> -----------------------------------------------------------
>
> (Updated Nov. 21, 2013, 11:18 p.m.)
>
>
> Review request for mesos, Benjamin Hindman, Ben Mahler, and Vinod Kone.
>
>
> Bugs: MESOS-823
> https://issues.apache.org/jira/browse/MESOS-823
>
>
> Repository: mesos-git
>
>
> Description
> -------
>
> See summary.
>
>
> Diffs
> -----
>
> src/tests/master_contender_detector_tests.cpp
> 5e4237454133edc155e74ffa04aec24ccd04c1b4
> src/tests/zookeeper_tests.cpp 0059438a26bfd03ee6da15029ca7a67674ea637e
>
> Diff: https://reviews.apache.org/r/15710/diff/
>
>
> Testing
> -------
>
> make check Linux (100 iterations) and OSX.
>
>
> Thanks,
>
> Jiang Yan Xu
>
>
