Hi Alan, I think you are right that CMAKE_LD_FLAGS has never done anything, and setting it was always a mistake.
The uses of CMAKE_LD_FLAGS in JNIFlags.cmake are harmless, since the -m32 option is not needed for the linker. However, it's more concerning that hadoop-mapreduce-client-nativetask seems to be using this macro to set "-no-undefined -version-info 0:1:0" and some other stuff. Can you create a JIRA to change these uses over to the correct CMake directive? If you create a JIRA that makes only that change, it should be pretty quick to review and commit. best, Colin On Thu, Jun 4, 2015 at 7:41 AM, Alan Burlison <alan.burli...@oracle.com> wrote: > On 03/06/2015 21:23, Alan Burlison wrote: > >> hadoop-common-project/hadoop-common/src/JNIFlags.cmake sets the >> CMAKE_LD_FLAGS variable to the flags required for the linker. However, >> despite the name, CMAKE_LD_FLAGS is *not* a valid standard CMake >> variable > > > Confirmed, I've put nonsense flags in CMAKE_LD_FLAGS that should produce an > invalid linker command-line and they don't appear in the generated makefiles > and the build works fine. If I put the same nonsense flags in the proper > variables then the build blows up as I'd expect. > > -- > Alan Burlison > --