1. 28 Sep, 2012 10 commits
  2. 27 Sep, 2012 11 commits
  3. 26 Sep, 2012 3 commits
    • Utkarsh Ayachit's avatar
      Added support for testing streaming. · 7de86c89
      Utkarsh Ayachit authored
      Added a new test to test AMR outline streaming. Currently only builtin test
      is supported. Need to extend smTestDriver for streaming to add client-server
      Change-Id: Ib6ed13b4c205f5c065bd9f9118c07d3e7ec962dd
    • Utkarsh Ayachit's avatar
    • Utkarsh Ayachit's avatar
      Fixed client-server AMR volume rendering issue. · a3709ae7
      Utkarsh Ayachit authored
      When locally rendering, the resampled volume was getting reset when a streamed
      piece was updated. This was because the
      vtkResampledAMRImageSource::UpdateResampledVolume() was checking the AMR data's
      mtime to determine if we needed to reinitialize the datastructure. That's
      incorrect. Since in client-server mode the client is delivered the AMR for every
      update, we end up with a new mtime and hence we were always reinitializing the
      datastructure. Fixed that. vtkAMRStreamingVolumeRepresentation::RequestData()
      takes care of initializing the vtkResampledAMRImageSource when the data changes
  4. 25 Sep, 2012 5 commits
    • Utkarsh Ayachit's avatar
      Fixing single-server remote-rendering configuration. · a0860760
      Utkarsh Ayachit authored
      AMR volume rendering now works for single-server, remote rendering use-case. We
      currently don't support local-rendering or multi-server configurations for AMR
      volume rendering.
    • Utkarsh Ayachit's avatar
      Remove old AMR volume rendering classes. · 2823e752
      Utkarsh Ayachit authored
      This removes some of the functionality previously present (like optimal
        resolution picking), but we need to revamp that code to work within the
      streaming framework anyways hence discarding these old classes.
    • Andrew Bauer's avatar
      Better way to add coprocessing test labels. · 3930eecd
      Andrew Bauer authored
      Change-Id: Ia05ac921125f116b92051291d5e5512c5a500f98
    • Andrew Bauer's avatar
      Giving coprocessing tests PARAVIEW and CATALYST labels. · 76e2ef44
      Andrew Bauer authored
      This will allow them to be run on the dashboards.
    • Andrew Bauer's avatar
      Fixing up coprocessing's FortranAdaptor. · b1fca474
      Andrew Bauer authored
      3 changes:
      1. Properly initializing vtkCPProcessor.
         vtkCPProcessor needed to be initialized before adding in a
      2. Exporting API methods so that they can be used by other libraries/executables.
         For GCC and Windows, the export macros weren't being used so that other
         libraries and executables that tried to link to methods in this library
         would cause problems.
      3. Change multiblock data set iterator calls to match with changes to VTK.
      Change-Id: I2ff2a7cb038989a8a42efc545f93b81dc7a69710
  5. 24 Sep, 2012 3 commits
    • Burlen Loring's avatar
      SciberQuestToolKit -- intel includes · 074cd488
      Burlen Loring authored
      gcc intinsics are erroneously found by the intel compiler, try to
      prevent this by finding and including intel's directory explicitly.
      if we can't locate the header then a warning is issued, the user
      may have to set the path if there are problems. There may not be
      problems if the intel compiler searches its directories bofore
      gcc's or gcc's copy of the intriniscs don't conflict with intels.
    • Utkarsh Ayachit's avatar
      BUG #13036. Fix Calculator panel. · 4c81d438
      Utkarsh Ayachit authored
      Coverting Calculator panel to new style to overcome bugs in the panel code.
      Adding a new pqCalculatorWidget which is a pqPropertyWidget that can be
      connected to any property that needs a calculator like functionality.
    • Utkarsh Ayachit's avatar
      Updating documentation. · 7723cdc1
      Utkarsh Ayachit authored
  6. 23 Sep, 2012 1 commit
    • Utkarsh Ayachit's avatar
      Fix initialization for executables that don't run with mpi. · 8f2346da
      Utkarsh Ayachit authored
      Processes that are not expected to be run with mpirun/mpiexec should use
      PROCESS_CLIENT and not PROCESS_BATCH. That avoids the MPI_Init() call and thus
      the errors on some mpi implementations where mpiexec must be used for such
      Change-Id: I6fe7f56e8acae5795154ffbe832186bd8198bf7c
  7. 22 Sep, 2012 7 commits