> On May 10, 2017, 9:19 p.m., Jeff Coffler wrote: > > support/windows-build.bat > > Line 52 (original), 52 (patched) > > <https://reviews.apache.org/r/59116/diff/1/?file=1714224#file1714224line52> > > > > Is this correct? Did we deprecate older toolsets? I thought that, at > > this point, we REQUIRE the newer version of Visual Studio.
This is correct. We do not yet require, only warn that it is deprecated; and this commit does not change the default. If it did, I'd be forcing Joe to update Jenkins, but I'm not (yet anyway). > On May 10, 2017, 9:19 p.m., Jeff Coffler wrote: > > support/windows-build.bat > > Line 61 (original) > > <https://reviews.apache.org/r/59116/diff/1/?file=1714224#file1714224line62> > > > > Since we're shifting from msbuild to cmake to build, does cmake assume > > /m (parallel builds)? Yes. > On May 10, 2017, 9:19 p.m., Jeff Coffler wrote: > > support/windows-build.bat > > Line 68 (original), 57 (patched) > > <https://reviews.apache.org/r/59116/diff/1/?file=1714224#file1714224line69> > > > > I know that we don't support NOT building Debug right now, but > > shouldn't this be configurable for windows-build.bat, perhaps with the > > default to build with debug? We don't support it yet, so I didn't add it as a broken option. > On May 10, 2017, 9:19 p.m., Jeff Coffler wrote: > > support/windows-build.bat > > Line 83 (original), 72 (patched) > > <https://reviews.apache.org/r/59116/diff/1/?file=1714224#file1714224line84> > > > > So this is actually different from Linux. > > > > On Linux, if you do a `make check`, libprocess, stout, and mesos tests > > are all run (even if one of them has an error). But on Windows with > > windows-build.bat, that's not the case. > > > > I think it should be. Perhaps save the error state so you can report > > (at the end) that libprocess and/or stout tests failed. But given that it's > > routine for one or more tests to fail today, it's annoying that > > windows-build.bat isn't as good as Linux in this regard. Totally agree, but wasn't fixing that here. It's a separate issue (and would require more careful testing of the script). - Andrew ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/59116/#review174560 ----------------------------------------------------------- On May 10, 2017, 8:40 p.m., Andrew Schwartzmeyer wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/59116/ > ----------------------------------------------------------- > > (Updated May 10, 2017, 8:40 p.m.) > > > Review request for mesos, Jeff Coffler, John Kordich, Joseph Wu, and Li Li. > > > Bugs: MESOS-7343 > https://issues.apache.org/jira/browse/MESOS-7343 > > > Repository: mesos > > > Description > ------- > > It now understands the environment variable `CMAKE_GENERATOR` so that > users can change the generator; the default is still Visual Studio 14. > > Direct usage of `msbuild` was replaced with `cmake --build`. > > We now build `mesos-tests` in the same pattern as `stout-tests` and > `libprocess-tests`. If admin priveleges are missing, we skip running the > tests, but still proceed to build the general target. > > We set the CMake toolset instead of an environment variable for > `PreferredToolArchitecture`. > > > Diffs > ----- > > support/windows-build.bat 88e177ced912b61f8169b58a50d1cf6ba0c4c0c7 > > > Diff: https://reviews.apache.org/r/59116/diff/1/ > > > Testing > ------- > > Built and tested script with clean build on Windows. > > > Thanks, > > Andrew Schwartzmeyer > >