> On April 14, 2014, 5:19 p.m., Benjamin Hindman wrote: > > configure.ac, lines 165-167 > > <https://reviews.apache.org/r/19487/diff/7/?file=555431#file555431line165> > > > > Given these changes, I feel like taking JAVA_LDFLAGS is misleading > > since we're not actually linking libjvm into anything (IIUC). In fact, I'd > > even prefer to simplify our configure tests to just compile (is there any > > advantage in checking that we can link?).
Correct, it will be removed in the next patch. - Timothy ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/19487/#review40273 ----------------------------------------------------------- On April 11, 2014, 8:28 p.m., Timothy St. Clair wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/19487/ > ----------------------------------------------------------- > > (Updated April 11, 2014, 8:28 p.m.) > > > Review request for mesos, Benjamin Hindman, Niklas Nielsen, and Till > Toenshoff. > > > Bugs: MESOS-1126 > https://issues.apache.org/jira/browse/MESOS-1126 > > > Repository: mesos-git > > > Description > ------- > > The dependency on libjvm is a "hard", requiring an link and rpath. This > causes issues when shifting between jvms. The standard practice is to use > dlopen against the $JAVA_JVM_LIBRARY location. > > MESOS-1126 > + Enable dlopen for libjvm.so > + Cleanup redundant jni inclusions > > This should make MESOS-855 a non-issue. > > > Diffs > ----- > > configure.ac c1de6d7 > src/Makefile.am 95f133d > src/common/build.hpp 115e9cc > src/common/build.cpp eed08ed > src/jvm/jvm.cpp d3dfa75 > > Diff: https://reviews.apache.org/r/19487/diff/ > > > Testing > ------- > > make check (across both Mac and Linux) > > > Thanks, > > Timothy St. Clair > >
