1. 16 Apr, 2018 4 commits
    • Burlen Loring's avatar
      support multiple meshes · 039f2281
      Burlen Loring authored
      Adds mesh name (string) parameter to data access methods in the data
      adaptor data API. Adds methods to query number of meshes and mesh names
      by index to data adaptor metadata API. Updates data adptors, analysis
      adaptors, and mini-apps too use the new API.
      Co-Authored-By: default avatarBrad Whitlock <bjw.ilight@gmail.com>
    • Burlen Loring's avatar
      do not SWIG -threads · d80a0f86
      Burlen Loring authored
      SWIG's -threads flag tells SWIG to release GIL before entering
      all wrapped functions, the idea being that Python can run concurently
      with the wrapped C++ code. This is only safe when the wrapped code
      aquires the GIL before making calls to the Python API, and the
      interpreter has to be initialized for threading. Niether of these
      are true by default in ParaView leading to mysterious segv's.
      Note, there is a CMake flag that can toggle the correct behavior
      at compile time. However, it is disabled by default so we will
      not use -threads.
    • Burlen Loring's avatar
    • Burlen Loring's avatar
      add mesh metadata · 2c2b82cf
      Burlen Loring authored
      Add a MeshMetadata class to organize and manage metadata associate
      with meshes such as mesh names, point and cell data arrays and ghost
      node and zone metadata.
      Co-Authored-By: default avatarBrad Whitlock <bjw.ilight@gmail.com>
  2. 13 Apr, 2018 4 commits
    • Burlen Loring's avatar
      add VTK utils · 2026a6ba
      Burlen Loring authored
      A collection of helper functions for common tasks dealing
      with VTK
    • Burlen Loring's avatar
      add data requirements · bbcd3d8c
      Burlen Loring authored
      a helper class to handle XML parsing and track mesh name,
      and point and cell data array names to be processed during
    • Burlen Loring's avatar
      analysis adaptor finalize api · 8cda178f
      Burlen Loring authored
      Add a finalize method, this should be called by the bridge
      before the analysis is deleted giving the analysis a chance
      to clean up and shut down. The motivation is to ease use of
      smart pointers with analyses that need to do MPI calls to
      clean up and shut down. Without an explicit finalize step
      the analysis potentially end up calling MPI after MPI_Finalize
      because the smart pointer goes out of scope after MPI_Finalize
      is called causing run to abort/crash.
    • Burlen Loring's avatar
      Document senseiADIOS::Schema · 4d00ff5c
      Burlen Loring authored
  3. 29 Jan, 2018 3 commits
  4. 25 Jan, 2018 4 commits
  5. 22 Jan, 2018 1 commit
  6. 19 Dec, 2017 7 commits
  7. 05 Dec, 2017 1 commit
  8. 04 Dec, 2017 1 commit
  9. 01 Dec, 2017 1 commit
  10. 30 Nov, 2017 5 commits
  11. 16 Nov, 2017 2 commits
  12. 08 Nov, 2017 7 commits