> On Sept. 30, 2016, 12:37 a.m., Joseph Wu wrote: > > 3rdparty/libprocess/include/process/pid_group.hpp, lines 54-59 > > <https://reviews.apache.org/r/50493/diff/3/?file=1464075#file1464075line54> > > > > We shouldn't have two ways of specifying the `base` PIDs. > > > > Let's remove the `base` from the constructor in favor of runtime > > initialization.
Fixed in https://reviews.apache.org/r/53700/ Somehow when I make extra patches in a review chain, those extras are branched off instead of merged in the same chain... > On Sept. 30, 2016, 12:37 a.m., Joseph Wu wrote: > > 3rdparty/libprocess/include/process/pid_group.hpp, line 59 > > <https://reviews.apache.org/r/50493/diff/3/?file=1464075#file1464075line59> > > > > Does this need to be a virtual function? > > > > e.g. Is there anything the module needs to do here that can't be done > > during construction or while being used by the replicated log? Different module implementations may do different things in `initialize()` and this gives freedom to module authors. > On Sept. 30, 2016, 12:37 a.m., Joseph Wu wrote: > > 3rdparty/libprocess/src/pid_group.cpp, lines 182-185 > > <https://reviews.apache.org/r/50493/diff/3/?file=1464076#file1464076line182> > > > > This shouldn't be a no-op (especially based on the above). changed to `base = {_base}` - Jay ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/50493/#review150971 ----------------------------------------------------------- On Nov. 14, 2016, 3:33 a.m., Jay Guo wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/50493/ > ----------------------------------------------------------- > > (Updated Nov. 14, 2016, 3:33 a.m.) > > > Review request for mesos and Joseph Wu. > > > Bugs: MESOS-5828 > https://issues.apache.org/jira/browse/MESOS-5828 > > > Repository: mesos > > > Description > ------- > > All subclasses derived from PIDGroup are required to have `base` > as a set of PIDs that always exist from the beginning. Therefore > `base` is refactored into PIDGroup base class. This is a step > towards modularizing replicated_log's PIDGroup abstraction. > > > Diffs > ----- > > 3rdparty/libprocess/include/process/pid_group.hpp PRE-CREATION > 3rdparty/libprocess/src/pid_group.cpp PRE-CREATION > > Diff: https://reviews.apache.org/r/50493/diff/ > > > Testing > ------- > > see end of review chain > > > Thanks, > > Jay Guo > >
