> On Sept. 23, 2013, 11:04 p.m., Ben Mahler wrote: > > 3rdparty/libprocess/3rdparty/stout/include/stout/duration.hpp, lines 286-287 > > <https://reviews.apache.org/r/11331/diff/4/?file=356211#file356211line286> > > > > What about making this class a friend of Duration (to gain access to > > the constants) and then instead of using casting with doubles you can use > > just use mod operations on integers: > > > > int64_t nanoseconds = duration.ns(); > > > > ... > > > > if (nanoseconds % MILLISECONDS != 0 && nanoseconds % MICROSECONDS > > == 0) { > > stream << duration.us() << "us"; > > } else { > > stream << duration.ms() << "ms"; > > } > > > > Would be nice to get rid of the double casting, thoughts? > > Jiang Yan Xu wrote: > OK it does look a bit better. If we trust compiler optimization and use > qualified names: > > if (duration.ns() % Duration::MILLISECONDS != 0 && > duration.ns() % Duration::MICROSECONDS == 0) { > stream << duration.us() << "us"; > } else { > stream << duration.ms() << "ms"; > } > > Looks OK? > > > > > Ben Mahler wrote: > Yep, why are compiler optimizations relevant here?
Oh, for avoiding re-computation of duration.ns()..? Let's not worry about that :) - Ben ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/11331/#review26328 ----------------------------------------------------------- On Sept. 23, 2013, 9:28 p.m., Jiang Yan Xu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/11331/ > ----------------------------------------------------------- > > (Updated Sept. 23, 2013, 9:28 p.m.) > > > Review request for mesos, Benjamin Hindman, Ben Mahler, Ian Downes, Jie Yu, > and Vinod Kone. > > > Bugs: MESOS-477 > https://issues.apache.org/jira/browse/MESOS-477 > > > Repository: mesos > > > Description > ------- > > - Currently when the result isn't a whole number, we go one-level down in > terms of the unit of time and see if we can get a whole number there. If so, > use the lower unit; if not, use the higher unit. > > > Diffs > ----- > > 3rdparty/libprocess/3rdparty/stout/include/stout/duration.hpp > 47e85ff8bbb310475dea382bfca9389c4295cf77 > 3rdparty/libprocess/3rdparty/stout/tests/duration_tests.cpp > e0910d4c82c2289f8d08deadbf4a8ef0448d0f76 > > Diff: https://reviews.apache.org/r/11331/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Jiang Yan Xu > >
