Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-6133

mpi4py does not compile under Yosemite due to hardcoded MACOSX_DEPLOYMENT_TARGET

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: lsst_distrib
    • Labels:
      None
    • Story Points:
      1
    • Epic Link:
    • Sprint:
      DRP F16-3
    • Team:
      Data Release Production

      Description

      mpi4py build on Yosemite (Mac OS X 10.10) fails with

      _build.log:[2016-05-17T16:51:55.847161Z] error: $MACOSX_DEPLOYMENT_TARGET mismatch: now "10.9" but "10.10" during configure
      

      For details see attached build log.

      The MACOSX_DEPLOYMENT_TARGET is being set in ups/eupspkg.cfg.sh

      [serenity mpi4py] cat ups/eupspkg.cfg.sh
      # If MACOSX_DEPLOYMENT_TARGET is not set, we force it to be at least 10.9
      # (Mavericks). This is the earliest version of OS X expected to work with
      # release 11 of the LSST stack.
      # This works around DM-5409, wherein mpi4py was attempting to use an OS X 10.5
      # SDK, based on querying Anaconda, and failing.
      export MACOSX_DEPLOYMENT_TARGET=${MACOSX_DEPLOYMENT_TARGET:-10.9}
      

      What is it that is supposed to be setting MACOSX_DEPLOYMENT_TARGET? And why is it not set at the time when ups/eupspkg.cfg.sh is run, but is set to 10.10 by the time the actually compilation is done?

        Attachments

          Issue Links

            Activity

            Hide
            wmwood-vasey Michael Wood-Vasey added a comment -

            Yes, I confirm that tickets/DM-6133 works for me

            [Calm lsstsw] rebuild -r tickets/DM-6133 mpi4py
                          mpi4py:  ok (1.0 sec).
                          python:  ok (0.8 sec).
                             mpi:  ok (0.7 sec).
                           mpich:  ok (3.4 sec).
            # BUILD ID: b2174
                          python: 0.0.5 (already installed).
                           mpich: tickets.DM-6133-g0b28c09ad9 ....................................................................................................................................................................................................................................................................ok (574.9 sec).
                             mpi: 0.0.1+2 ok (4.0 sec).
                          mpi4py: tickets.DM-6133-gb8eebf5807 ok (36.0 sec).
            # BUILD b2174 completed.
            

            Show
            wmwood-vasey Michael Wood-Vasey added a comment - Yes, I confirm that tickets/ DM-6133 works for me [Calm lsstsw] rebuild -r tickets/DM-6133 mpi4py mpi4py: ok (1.0 sec). python: ok (0.8 sec). mpi: ok (0.7 sec). mpich: ok (3.4 sec). # BUILD ID: b2174 python: 0.0.5 (already installed). mpich: tickets.DM-6133-g0b28c09ad9 ....................................................................................................................................................................................................................................................................ok (574.9 sec). mpi: 0.0.1+2 ok (4.0 sec). mpi4py: tickets.DM-6133-gb8eebf5807 ok (36.0 sec). # BUILD b2174 completed.
            Hide
            swinbank John Swinbank added a comment -

            Michael Wood-Vasey – any objection to my deleting the tickets/DM-6133 branch on mpich before sending this for review?

            Show
            swinbank John Swinbank added a comment - Michael Wood-Vasey – any objection to my deleting the tickets/ DM-6133 branch on mpich before sending this for review?
            Hide
            swinbank John Swinbank added a comment -

            Tim Jenness, do you have time to look at this? It's ugly (suggestions as to how to make it less so welcome), but it seems to work.

            Note that you're only looking at the branch in mpi4py; ignore the changes in mpich (see my comment above).

            Show
            swinbank John Swinbank added a comment - Tim Jenness , do you have time to look at this? It's ugly (suggestions as to how to make it less so welcome), but it seems to work. Note that you're only looking at the branch in mpi4py ; ignore the changes in mpich (see my comment above).
            Hide
            tjenness Tim Jenness added a comment -

            Definitely insane. Go for it.

            Show
            tjenness Tim Jenness added a comment - Definitely insane. Go for it.
            Hide
            swinbank John Swinbank added a comment -

            Thanks; merged.

            Show
            swinbank John Swinbank added a comment - Thanks; merged.

              People

              Assignee:
              swinbank John Swinbank
              Reporter:
              wmwood-vasey Michael Wood-Vasey
              Reviewers:
              Tim Jenness
              Watchers:
              John Swinbank, Joshua Hoblitt, Michael Wood-Vasey, Tim Jenness
              Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

                Dates

                Created:
                Updated:
                Resolved:

                  Jenkins

                  No builds found.