On Sep 25, 2015, at 5:41 PM, Howard Pritchard <hpprit...@gmail.com> wrote: > > Well, jenkins doesn't lie.
Jenkins actually frequently *does* lie; that's kinda the problem... :-) > http://jenkins.open-mpi.org/job/ompi_master_cle5.2up02/595/console > > Looks like the commits associated with PR 595 busted mpi_f08 build. PR 595 is a libfabric-related PR from months ago. I'm guessing you mean https://github.com/open-mpi/ompi/commit/3698621df0bffd3711be2f2d81afd8e3ae2947e6; my commit from this morning. The Forum is over for today; sorry I wasn't available to fix it earlier. Just to be clear: the build didn't break (i.e., "make all" doesn't fail), meaning that this issue wasn't critical/time-sensitive to fix -- it could have waited until now for me to fix it (admittedly at the cost of several bot:retests for PRs from today). I tested with gcc and icc before pushing to master, but yes, I did miss that gcc longer built the mpi_f08 module. And somehow I missed that icc didn't install the mpi_f08 module -- my testing of the user's test program worked, so I must have goofed my testing somehow. Jet lag...? -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
