1. 11 Dec, 2018 1 commit
  2. 07 Dec, 2018 1 commit
  3. 05 Dec, 2018 1 commit
  4. 26 Nov, 2018 1 commit
    • Cory Quammen's avatar
      Change to breadth-first traversal when collapsing a KDTree · 559ecc4e
      Cory Quammen authored
      The MoveSubtree() function moves trees from lower down in the KDTree
      up. The existing depth-first traversal recursive algorithm in this
      function worked okay when descending the tree, but when moving back up
      the tree, elements in the array would be clobbered before they were
      moved to their correct destination. This would lead to some views
      being incorrectly deleted and moved into the wrong location when a
      level of the KDTree was being collapsed.
      
      Change to a breadth-first algorithm so that elements in the tree are
      always copied from lower to higher positions in the tree. This
      resolves the incorrect subtree movement.
      559ecc4e
  5. 20 Nov, 2018 1 commit
  6. 24 Oct, 2018 1 commit
  7. 16 Oct, 2018 1 commit
    • Mathieu Westphal's avatar
      Using StepColorSpace · 4b6fecc2
      Mathieu Westphal authored
      A new ColorSpace has been added in VTK, this uses it.
      This add a few Step Color Map preset
      This adds also .md and test
      4b6fecc2
  8. 27 Sep, 2018 1 commit
  9. 21 Sep, 2018 1 commit
    • David E. DeMarle's avatar
      export spatio temporal scripts · f38a6510
      David E. DeMarle authored
      two new pieces of global state:
       time compartment size, the number of ranks to use per timestep
       input filepattern for readers that begin the pipeline
      
      input filepattern used to be per source, and was specified via
      the channel matching section of the dialog. Per discussion with Andy
      multiple readers are not really well supported anyway so simplify
      to a global filepattern.
      
      fixed some bugs in the original plugin that this deprecates, involving
      a lookup into map of views that failed with no name, and code rot
      where the base paraview python API changed and made it incompatible
      f38a6510
  10. 07 Sep, 2018 1 commit
  11. 05 Sep, 2018 1 commit
  12. 04 Sep, 2018 1 commit
  13. 03 Sep, 2018 2 commits
    • Utkarsh Ayachit's avatar
      improve vtkPVDataDeliveryManager's data delivery logic. · 034e49d4
      Utkarsh Ayachit authored
      When doing remote rendering, the client's mtime for when data delivery
      happened was not correctly updated. That was causing the data delivery
      logic to reexecute on each render. That was causing #18340.
      
      Fixed vtkPVDataDeliveryManager to keep separate delivery objects for
      each delivery mode. This ensures that the client doesn't end up with a
      data delivery timestamp mismatch, as before. Also simplified the code in
      general and adding timer log entries to catch such bug easily in the
      future.
      
      Fixes #18340.
      034e49d4
    • Utkarsh Ayachit's avatar
      rename API to better reflect implementation. · 47089b5c
      Utkarsh Ayachit authored
      `vtkPVRenderView::GetUseDistributedRenderingForStillRender` and
      `vtkPVRenderView::GetUseDistributedRenderingForInteractiveRender` are
      renamed to better reflect the implementation. These are not for still vs
      interactive but full-res vs lod. New names avoid that confusion.
      47089b5c
  14. 29 Aug, 2018 2 commits
    • Utkarsh Ayachit's avatar
      cleanup spreadsheet view · 9fe63c47
      Utkarsh Ayachit authored
      This commit was motivated by a fix for BUG #17739. This includes the
      following improvements:
      
      1. added properties on SpreadSheetView to control aspects like hidden
         columns. There were two mechanisms at play to control the columns
         that are hidden one in the Qt part and another in the View for
         exporting. Combined those to use a single approach.
      2. Simplified pqSpreadSheetViewDecorator to update properties on the
         view for things like cell connectivity, field association etc. This
         make undo/redo, state save/restore work better for several of the
         parameters. A few still remain (see #18388).
      3. Changed CSV export to use column labels. Also made CSV export always
         only export visible columns. The toggle we previosly exposed doesn't
         make sense. Export capability is meant to export what is shown in the
         view. The option to export extra columns was unnecessary.
      4. Added a mechansim to vtkSMViewProxy subclasses to update their
         properties when representation are hidden/shown via the controller.
         vtkSMSpreadSheetViewProxy uses this mechanism to pick a good field
         association for the visible source.
      5. pqExportReaction was messing up undo/redo stack. Fixed that.
         pqExportReaction should not be affecting the undo/redo stack at all.
      
      (cherry picked from commit 4679f456)
      9fe63c47
    • David E. DeMarle's avatar
      Revert "cleanup spreadsheet view" · ae9b8a1f
      David E. DeMarle authored
      This reverts commit 4679f456.
      ae9b8a1f
  15. 23 Aug, 2018 1 commit
    • Utkarsh Ayachit's avatar
      cleanup spreadsheet view · 4679f456
      Utkarsh Ayachit authored
      This commit was motivated by a fix for BUG #17739. This includes the
      following improvements:
      
      1. added properties on SpreadSheetView to control aspects like hidden
         columns. There were two mechanisms at play to control the columns
         that are hidden one in the Qt part and another in the View for
         exporting. Combined those to use a single approach.
      2. Simplified pqSpreadSheetViewDecorator to update properties on the
         view for things like cell connectivity, field association etc. This
         make undo/redo, state save/restore work better for several of the
         parameters. A few still remain (see #18388).
      3. Changed CSV export to use column labels. Also made CSV export always
         only export visible columns. The toggle we previosly exposed doesn't
         make sense. Export capability is meant to export what is shown in the
         view. The option to export extra columns was unnecessary.
      4. Added a mechansim to vtkSMViewProxy subclasses to update their
         properties when representation are hidden/shown via the controller.
         vtkSMSpreadSheetViewProxy uses this mechanism to pick a good field
         association for the visible source.
      5. pqExportReaction was messing up undo/redo stack. Fixed that.
         pqExportReaction should not be affecting the undo/redo stack at all.
      4679f456
  16. 21 Aug, 2018 1 commit
    • David E. DeMarle's avatar
      new catalyst/cinema export GUI. · 8bbd0e75
      David E. DeMarle authored
      The idea is a central panel for the user to declare what they want to
      export, separated from the action of doing the export. The export
      configuration is saveable and restorable in state files. Combined these
      make it easier for the user to make Catalyst exports, avoid
      some annoying behaviors, and open up paths to some new features.
      
      The implementation uses savescreenshot and writerproxies instead of Catalyst
      specific export proxies. We use standard proxies rather than Catalyst specific
      ones to support new "export now" and "export batch script that will export"
      use cases. The proxies are managed in something called an ExportProxyDepot for
      more convenient access by the code.
      8bbd0e75
  17. 17 Aug, 2018 1 commit
  18. 03 Aug, 2018 2 commits
    • Utkarsh Ayachit's avatar
      fix vtkPVProminentValuesInformation. · 87bf513b
      Utkarsh Ayachit authored
      This fixes a few issues with how "prominent values" were handled.
      
      1. `vtkSMRepresentationProxy::GetProminentValuesInformation` always
         looked are represented data and never and it's input's data. That is
         different from how info for non-prominent values is handled.
         When coloring etc, we first look at the array range for the array
         from input, if present and then from represented data if the array
         was not present on the input. This was causing the molecule
         representation to not have any atomic numbers related information on
         creation. fixed that.
      
      2. `vtkPVProminentValuesInformation` didn't have correct code in
         `CopyFromObject`. It was just a copy-paste from vtkPVDataInformation
         which didn't make sense for vtkPVProminentValuesInformation. Fixed
         that.
      87bf513b
    • Utkarsh Ayachit's avatar
      Improve logic to mark representations modified. · cc766dc6
      Utkarsh Ayachit authored
      Improves the infrastructure to mark representations modified only when
      necessary. Current implementation (via
      vtkSMProxyProperty::SkipDependency) was error prone. Now,
      vtkSMRepresentationProxy is smarter at deciding is the proxy is being
      dirtied via a upstream pipeline connection or something else.
      
      This was done by adding new API (vtkSMProxy::MarkDirtyFromProducer)
      which has relevant information available to representation to determine
      which producer connection the dirty notification is coming from.
      
      This helps use avoid clearning animation cache for representation
      property changes that should only affect rendering (not data
      processing). Fixes #18275.
      
      Futhermore, all representations now have `DebugName` which is handy
      string that helps identify a representation proxy making it clear which
      source's what representation is corresponds to.
      
      vtkSMProxyProperty::SkipDependency (and `skip_dependency` XML attribute)
      is now obsolete and no longer needed or used.
      cc766dc6
  19. 01 Aug, 2018 1 commit
  20. 24 Jul, 2018 1 commit
    • Utkarsh Ayachit's avatar
      Fix issues with vtkPVRenderView::SetForceDataDistributionMode. · ce9f4e1d
      Utkarsh Ayachit authored
      When distribution mode was forced on vtkPVRenderView, we're operating in
      a non-standard environment where the nodes that ParaView things should
      have the geometry for rendering (based on the which nodes are doing the
      rendering) may not have the correct data. For example, to force all
      geometry rendering to happen on rank 0, a representation may set the
      forced data distribution mode to `vtkMPIMoveData::COLLECT` and still be
      in remote rendering mode (and not render-on-client). Since forced
      distribution mode can thrown a wrench in the logic to keep track of
      which data have been delivered to which nodes, we opt for being
      conservative and simply clear delivery metadata when forced render mode
      changes.
      
      This builds on fixes in 3b15cd74.
      ce9f4e1d
  21. 18 Jul, 2018 1 commit
  22. 17 Jul, 2018 2 commits
  23. 26 Jun, 2018 5 commits
  24. 24 Jun, 2018 2 commits
  25. 16 Jun, 2018 2 commits
  26. 11 Jun, 2018 1 commit
    • Allison Vacanti's avatar
      Limit the rescale factor to keep it sane. · a911421d
      Allison Vacanti authored
      In issue #18220, a user was saving a screenshot scaled from
      {400, 400} --> {1641, 1432}, and it ended up resizing the
      screen to 3 px wide with a magnification factor of 547 in
      the X dimension. Eventually this led to trying to render
      text at 39384 dpi, which got real ugly real quick.
      a911421d
  27. 24 May, 2018 1 commit
    • luz.paz's avatar
      Misc. typos · e012debf
      luz.paz authored
      Found via `codespell` and `grep`
      e012debf
  28. 25 Apr, 2018 1 commit
  29. 23 Mar, 2018 1 commit
  30. 28 Feb, 2018 1 commit
    • Michael Migliore's avatar
      Add a new checkbox in Volume representation to avoid confusion with MapScalars · 2352686f
      Michael Migliore authored
      When volume representation is selected and **Map Scalars** and selected data
      array has 2 or 4 components, a new property called **Mutliple Components Mapping**
      is available.
      
      When this new feature is enabled:
      * for a two-components array, the first component is mapped to the color, the
        second component is mapped to the opacity.
      * for a four-components array, the first, second and third components are used
        as RGB values and the fourth component is mapped to the opacity.
      
      Note that:
      * this feature is only available when **MapScalars** is *ON* ;
      * this feature forces **Use Separate Color Map** ;
      * it scales the color and the opacity range with the correct component.
      2352686f