> On April 13, 2016, 9:27 a.m., Alexander Rukletsov wrote: > > src/tests/master_allocator_tests.cpp, line 1604 > > <https://reviews.apache.org/r/46135/diff/1/?file=1342280#file1342280line1604> > > > > If you're not doing any checks with resources, you can use > > `Future<Nothing>` instead. > > Yongqiao Wang wrote: > They are different variable type, 'const mesos::Resources' can not > convert to 'const Nothing'.
You use `Future<Nothing>` together with `FutureSatisfy`. The difference between `FutureArg` and `FutureSatisfy` is that the latter sets the future when the method is invoked, while the former _additionally_ saves the actual value of some argument. If you don't use this value later on, you don't have to use `FutureArg`. > On April 13, 2016, 9:27 a.m., Alexander Rukletsov wrote: > > src/tests/master_allocator_tests.cpp, line 1637 > > <https://reviews.apache.org/r/46135/diff/1/?file=1342280#file1342280line1637> > > > > You don't need this `settle()`. > > Yongqiao Wang wrote: > According to my understanding, call Clock::settle() after > Clock::advance(), then it can ensure allocate() is executed before do the > following check. So I think it is needed. `AWAIT_READY` does that for you if the clock is paused. Quoting BenM: "Think about it this way, if you put a `settle()` before the `AWAIT_READY`, then the implication is that you don't need `AWAIT_READY` at all, you could just do `ASSERT_TRUE(f.isReady())` because you've done the necessary settling explicitly. This is why it seems strange to have a `settle` immediately preceed an `AWAIT_READY`, its the job of `AWAIT_READY` to wait until the future is ready (which means doing the right thing when the clock is paused, otherwise we would have to write a lot of `settle` calls)." - Alexander ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/46135/#review128640 ----------------------------------------------------------- On April 13, 2016, 3:21 a.m., Yongqiao Wang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/46135/ > ----------------------------------------------------------- > > (Updated April 13, 2016, 3:21 a.m.) > > > Review request for mesos, Adam B and Alexander Rukletsov. > > > Bugs: MESOS-5146 > https://issues.apache.org/jira/browse/MESOS-5146 > > > Repository: mesos > > > Description > ------- > > Pause the clock and make sure all resources from rescinded > offers are recovered before doing a batch allocation. > > > Diffs > ----- > > src/tests/master_allocator_tests.cpp > 404ff098baf89bf2a1c6e32424d591a6ea1a093c > > Diff: https://reviews.apache.org/r/46135/diff/ > > > Testing > ------- > > Make && Make check. > > ``` > Yongs-MacBook-Pro:build yqwyq$ ./src/mesos-tests > --gtest_filter=MasterAllocatorTest/0.RebalancedForUpdatedWeights > [==========] Running 1 test from 1 test case. > [----------] Global test environment set-up. > [----------] 1 test from MasterAllocatorTest/0, where TypeParam = > mesos::internal::master::allocator::MesosAllocator<mesos::internal::master::allocator::HierarchicalAllocatorProcess<mesos::internal::master::allocator::DRFSorter, > mesos::internal::master::allocator::DRFSorter, > mesos::internal::master::allocator::DRFSorter> > > [ RUN ] MasterAllocatorTest/0.RebalancedForUpdatedWeights > [ OK ] MasterAllocatorTest/0.RebalancedForUpdatedWeights (385 ms) > [----------] 1 test from MasterAllocatorTest/0 (386 ms total) > > [----------] Global test environment tear-down > [==========] 1 test from 1 test case ran. (396 ms total) > [ PASSED ] 1 test. > > ``` > > > Thanks, > > Yongqiao Wang > >
