Minutes: Minutes for 22 January 2016

Fri 22 January 2016
By OpenCMISS Project.

Present: Chris, Noel, Prasad, Daniel, Soroush, Hashem, Hugh, Andre

Apologies: Richard

  1. Updates

    • Prasad: getting building set up, aiming at contact mechanics

      tidy-up; looking at using python bindings and how to set that up. Reviving some old code from a really old version of OpenCMISS, useful code will get merged into current devel. This will get Nancy’s work mostly back into the prime repo.

    • Daniel: windows CMake builds (see below), cmake build working on a

      range of machines with feature tests passing, windows building but feature tests not running (hopefully trivial fixes), VS 2013 zinc build (debug and release). Can try to get windows intel compiler license to try iron builds in VS. Using intel compilers not MS compilers. Need to overhaul the symbols dependencies are exporting (cmake PUBLIC, PRIVATE, INTERFACE).

    • Soroush: implemented new geometry from brazilian group. 10 hours

      to simulate 1 time period on one processor. Going to multiple processors (2+) resulted in mismatches in the connectivity between regions on different processors. Now relearning parallel debugging. Element based flow direction now implemented. Implementing acid-base physiology CellML models.

    • Hashem: Imaging and image processing to get data to use in fitting

      geometric meshes. Host mesh fitting developments in own fork, need to get merged back to prime/devel. Working on framework for predicting growth/growth models. Cubic hermite is the way to go for future use in mechanics and growth. Prasad suggesting considering alternative elements eg cubic lagrange for simplicity.

    • Hugh: Updated github repo for Zinc - now to be used as the prime

      zinc repo. SVN repo should be deleted next week before Richard is back (perhaps just switched to readonly) - will try to get developers to stop pushing changes into the svn repo using a memo for now. Zinc source and tests moved into github repo, bindings next on the list. Getting instruction from Marty and Prasad as to the requirements for Neon for the 3rd year mechanics class (321) - putting together plan now, time to get things changed before implementing. Noel currently generating documentation from the SVN repo, but can switch over to github repo.

    • Andre: hopefully fixed windows bug with OpenCMISS cellml (cellml

      component in the cmake system).

    • Chris: Looking at cmake build on many platforms - seems to be

      working, iterating on some issues with Daniel and getting through them. Building on NeSI with Intel MKL with old code from Nancy and mostly got that working. Need to check on cmake build on NeSI to see if it worked (default build).

    • Noel: minutes section added to the website (currently listed on

      front page, but will remove that or move it to a developer’s page). Website issues to be worked on: documentation landing page to be moved into the sphinx project for better consistency of user experience; supporting documentation versioning.

  2. Proposal for documentation versions on the website

    • opencmiss.org/documentation/ -
      • landing page, version selector (e.g. cmake.org/documentation);
      or
      • latest release (e.g. Chris had an example but I can’t recall it).

      Discussion on which of the above is best, no agreement achieved today. Think about it and we can make a decision next week.

    • opencmiss.org/documentation/v1.1/ - documentation for release 1.1

    • opencmiss.org/documentation/v1.2 …

    • opencmiss.org/documentation/devel - development version docs

  3. Status of Windows builds

    • priority on VS or mingw? can have a quick look at using Intel

      compilers with VS and Iron to see if achievable in the time remaining with Daniel here.

    • default config for building with VS on windows? need a fortran

      compiler in order to build Iron but not Zinc. Default should be to give an error if no fortran compiler available and user required to read the documentation and turn off Iron if they still want to build. For consistency should have similar failure for all components - if they are turned on and something is missing that is an error.

  4. Examples status

    • OpenCMISS/examples - all the examples in one repo

    • OpenCMISS-Examples/examples - all the examples as submodules to

      match tree from OpenCMISS/examples, each example in its own repo, Daniel has a process to update things from OpenCMISS/examples. This is mostly used for maintenance.

    • Cool python script to copy/update examples from OpenCMISS/examples

      to individual repositories under OpenCMISS-Examples and update submodule links in OpenCMISS-Examples/examples.

    • Need to confirm state of all examples under old Makefile build

      system. Will give a base for determining when the CMake build system replicates same functionality as Makefile build system.

    • Need to notify people to submit back any outstanding changes to

      OpenCMISS/examples and let them know that the repo will disappear shortly in favour of individual repos under OpenCMISS-Examples GitHub organisation.

  5. Python bindings

    • multiple architecture/configs installs of python bindings

    • Chris doesn’t like the use of virtual environments for the “dumb

      user”, believes it to be more complex than a user could cope with.

    • Need to investigate possible alternatives to see if there is any

      other way to manage python environments.