> On March 22, 2016, 9:52 p.m., Neil Conway wrote: > > 3rdparty/libprocess/include/process/logging.hpp, line 57 > > <https://reviews.apache.org/r/44135/diff/2/?file=1306311#file1306311line57> > > > > I wonder whether we can use something from C++11 to avoid the need for > > non-portable code here. From a quick skim of the reference page, > > `std::atomic_thread_fence(std::memory_order_seq_cst)` _might_ work, but I > > haven't dug in enough to say for sure.
It may be posible, in fact std::atomic_thread_fence uses MemoryBarrier on x64. However this requires glog patching. - Daniel ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/44135/#review124898 ----------------------------------------------------------- On March 18, 2016, 6:49 p.m., Daniel Pravat wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/44135/ > ----------------------------------------------------------- > > (Updated March 18, 2016, 6:49 p.m.) > > > Review request for mesos, Alex Naparu, Daniel Pravat, Artem Harutyunyan, Alex > Clemmer, Joris Van Remoortere, Michael Park, M Lawindi, and Yi Sun. > > > Repository: mesos > > > Description > ------- > > Libprocess: Use Windows-compatible memory fence in logging. > > > Diffs > ----- > > 3rdparty/libprocess/include/process/logging.hpp > ee680ae8ecd837f24ec7731d55680ad8440808db > > Diff: https://reviews.apache.org/r/44135/diff/ > > > Testing > ------- > > OSX: make > Windows: build/run > > > Thanks, > > Daniel Pravat > >
