> On Feb. 25, 2016, 12:22 a.m., Michael Park wrote: > > 3rdparty/libprocess/3rdparty/stout/include/stout/abort.hpp, line 52 > > <https://reviews.apache.org/r/43416/diff/2/?file=1253414#file1253414line52> > > > > Why is the cast to `size_t` needed here? > > Alex Clemmer wrote: > It is required, and (though it has been awhile) I believe it's because on > 64-bit machines, without this type information, 1 could be one of several > integral types. The cast disambiguates. > > Michael Park wrote: > I only see one definition of `::write` in > `3rdparty/libprocess/3rdparty/stout/include/stout/windows.hpp`, which has the > signature: `inline auto write(int fd, const void* buffer, size_t count)`. > > Doesn't seem like there should be any ambiguity here, are there other > defintions of `write` that we're contending with? Because `1` is clearly > convertible to `size_t`. > > Alex Clemmer wrote: > There is also `::write` in the Windows implementation of the C Runtime. > To confirm that this is the problem, we can delete the `inline auto write` > you mention, and the compile problem disappears. (I just confirmed this is > true.) > > So, practically speaking, there seem to be two choices: > > (1) Delete the `::write` in windows.hpp. > (2) Add a cast here. > > Note that if we choose (1) we should probably then fork every file where > where we call `::write` in the code, and replace those calls with `::_write`, > as the former is deprecated.
I added a comment to the review for posterity. - Alex ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/43416/#review120623 ----------------------------------------------------------- On Feb. 26, 2016, 9:23 p.m., Alex Clemmer wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/43416/ > ----------------------------------------------------------- > > (Updated Feb. 26, 2016, 9:23 p.m.) > > > Review request for mesos, Alex Naparu, Daniel Pravat, Artem Harutyunyan, > Joris Van Remoortere, Michael Park, M Lawindi, and Yi Sun. > > > Repository: mesos > > > Description > ------- > > Windows: Removed ambiguous call to `::write`. > > > Diffs > ----- > > 3rdparty/libprocess/3rdparty/stout/include/stout/abort.hpp > 88b355e09f76f0412c74ad69556572f0079deb8f > > Diff: https://reviews.apache.org/r/43416/diff/ > > > Testing > ------- > > > Thanks, > > Alex Clemmer > >
