> On April 1, 2014, 8:50 p.m., Benjamin Hindman wrote: > > src/Makefile.am, lines 161-163 > > <https://reviews.apache.org/r/19854/diff/2/?file=543618#file543618line161> > > > > I suggested pulling out the offending translation units so that people > > could have some context to understand what you mean by this comment. Maybe > > I'm just the ignorant one, but I'd like this comment to be even more > > explicit. Here is another suggestion: > > > > ---- > > As of automake 1.14 the 'subdir-objects' option is turned on by default > > and so we can't have the same file names in sub-directories because the > > resulting object files collide (... how do the relative paths play a role > > here?). For example, we can't have both a master/constants.cpp and a > > slave/constants.cpp. > > ---- > > > > Also, is this fixable? Can we add a TODO and include suggestions to fix > > this so that someone else doesn't need to try and think of those > > suggestions themselves in the future? > > Timothy St. Clair wrote: > I decided to dig deeper into this and uncovered that my original version > of 1.14 was a pre-release build, and quite buggy. > http://koji.fedoraproject.org/koji/buildinfo?buildID=470561. In that build, > my object files were colliding even though they should not. Upon update to > automake-1.14.1-1 the issue does not exist, so I've removed the modification. > > Benjamin Hindman wrote: > Oh awesome. Are we sure that the VPATH bug is not still a problem though? > See http://lists.gnu.org/archive/html/automake-patches/2013-06/msg00011.html. > I thought that was the issue from MESOS-577. Perhaps that's been fixed in > automake 1.14.1 but anyone with 1.14 would still have this issue? > > Timothy St. Clair wrote: > That particular issue does not seem to appear in my env, from what I see. > > > find . | grep "\\$" -> nothing > ran `make check` and it comes out green. > > Do you have a specific example of how it failed? I could try to repro. > > Also, the package maintainers who track, have already shifted in both RH > & Deb channels before 1.14 has gotten into the wild. > > https://packages.debian.org/search?keywords=automake > http://koji.fedoraproject.org/koji/packageinfo?packageID=1087 > It also appears that MacPorts has been updated too.
I think the failure occurred on OS X, but given that you tried this out on the latest OS X I'm happy to move forward. Let's keep an eye out to see if this crops up again so we can gate 1.14 if it has issues even though 1.14.1 works. Thanks Tim! - Benjamin ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/19854/#review39201 ----------------------------------------------------------- On April 3, 2014, 2:31 p.m., Timothy St. Clair wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/19854/ > ----------------------------------------------------------- > > (Updated April 3, 2014, 2:31 p.m.) > > > Review request for mesos and Benjamin Hindman. > > > Bugs: MESOS-577 > https://issues.apache.org/jira/browse/MESOS-577 > > > Repository: mesos-git > > > Description > ------- > > Enable automake 1.14 > > > Diffs > ----- > > 3rdparty/libprocess/configure.ac 2e0b72c > bootstrap ed0bc36 > configure.ac 5404dc2 > > Diff: https://reviews.apache.org/r/19854/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Timothy St. Clair > >
