- Sep 25, 2024
-
-
a909b3af vtkDataAssemblyUtilities: Use vector based dataset indices insertion 45a5aa15 vtkAMRDataInternals: Improve performance when adding datasets Acked-by:
Kitware Robot <kwrobot@kitware.com> Tested-by:
buildbot <buildbot@kitware.com> Reviewed-by:
Cory Quammen <cory.quammen@kitware.com>
Merge-request: !11511
-
- Sep 24, 2024
-
-
b28170d4 catalyst: forward `CATALYST_WITH_EXTERNAL_CONDUIT` property Acked-by:
-
- Sep 23, 2024
-
-
Spiros Tsalikis authored
This avoids incremental checks against all existing indices, and instead checks are batched per amr level.
-
Spiros Tsalikis authored
Better complexity reduces the number of internal swaps needed.
-
This allows ParaView (and other projects) to use `VTK::catalyst` as `CATALYST_TARGET` when creating a Catalyst implementation. (cherry picked from commit 2a0fcba7)
-
- Sep 20, 2024
-
-
286fd622 Add a baseline again 8f929179 Remove incorrectly added file Acked-by:
-
- Sep 18, 2024
- Sep 16, 2024
-
-
8f81b6dd VRCollaboration: Do not close the same socket multiple times 2b2380bb Fix XML partitioned data writers to use proper subdirectory 2c077bd0 Fix array write beyond MaxId in vtkSMPMergePoints 0ac66bad Ensure atomicity of insertion Id 6c91bde1 Minor fix to vtkCleanPolyData when not merging 7ddd7d29 Minor fix to vtkPolyLineWidget representation c5f42c60 Fix loop count in vtkImageTracerWidget bd202800 Init the IntegrationTime array to number of points ... Acked-by:
Mathieu Westphal (Kitware) <mathieu.westphal@kitware.com> Reviewed-by:
Thomas Galland <thomas.galland@kitware.com> Merge-request: !11479
-
- Sep 10, 2024
-
-
In situations where the server is not actually present, multiple start/stop sequences of the collaboration client could cause seemingly randomly located aborts. By setting the socket pointers to nullptr after close them, we avoid closing them again when start is called, which seems to prevent the hard-to-predict zmq abort behavior. (cherry picked from commit 8ddcd34f)
-
The filename split was based on the first period instead of the last. If the user had multiple periods in the filename then the first period would be used as the separator between the base filename and the extension. For writing out file series this caused data to be overwritten since the same subdirectory was being used for each time step. (cherry picked from commit 77c64123)
-
Calling SetNumberOfPoints()/SetNumberOfTuples() on the points and arrays before merging the results ensures that we never index an array past the MaxId of the array. We call Resize() first, since Resize() will preserve existing data on reallocation, while SetNumberOfTuples() won't preserve data on reallocation, but will preserve data if the array stays the same size or shrinks. (cherry picked from commit 7e1f8ef7)
-
(cherry picked from commit 7c5e963e)
-
CallSetNumberOfPoints() at beginning rather than only allocating. This ensures that there is no point access beyond the MaxId of the point data array. The points array is shrunk to the final size after execution is complete. (cherry picked from commit e5199790)
-
Insert points to ensure points are properly sized. (cherry picked from commit 4e276502)
-
This is the loop to project the polyline points to a plane. (cherry picked from commit 007fe299)
-
This matches code elsewhere in FlowPaths, e.g. vtkStreamSurface.cxx. (cherry picked from commit 407dea94)
-
Allocate() just reserves memory, e.g. in preparation for insertion, but SetNumberOfTuples() is necessary to set size before SetTuple(). (cherry picked from commit 8cf2fc7c)
-
When vtkDataSetAttributes::CopyData() inserts a tuple, it will adjust MaxId if MaxId is less than the tuple index. We must ensure this doesn't happen when multithreading, or the threads will fight over MaxId. (cherry picked from commit a180e106)
-
Previously, when the ADIOS2 VTX reader read in most data arrays from the ADIOS2 file, it would silently leave a null array if that array did not exist. This opened up the likely consequence of the program later crashing when the reader attempted to use this array. Instead, the reader now reports an error when an array it attempts to read in is missing. This prevents subsequent problems. (cherry picked from commit 24959a33)
-
(cherry picked from commit 8a2b8305)
-
(cherry picked from commit 6ba1e023)
-
(cherry picked from commit a96d29eb)
-
(cherry picked from commit 47c85475)
-
When getting a naked pointer for writing to an array, the WritePointer() method ensures the array is adjusted so that it can receive the new values. If just GetPointer() is used, then the MaxId of the array is not correctly set. (cherry picked from commit 2ac14b21)
-
Not enough tuples were added to the arrays. (cherry picked from commit 022be287)
-
The Atom and Bond data arrays were not regenerated when the number of bonds and atoms was changed. (cherry picked from commit a851fd71)
-
This very old test was using arrays in strange ways, resulting in array access beyond the MaxId of the arrays. (cherry picked from commit f0cb53b7)
-
The size of the array was not being adjusted to the cell size prior to use. As a result, the array was being written to beyond MaxId, though still within the array's allocation. (cherry picked from commit 7b11a643)
-
The InsertComponent() method ensures that the MaxId of the array is adjusted, so that out-of-bounds accesses do not occur. (cherry picked from commit 9ef902e1)
-
The ghost arrays were built with SetValue() without adjusting the MaxId to indicate the number of values in the array. With InsertValue(), the MaxId is adjusted automatically. (cherry picked from commit e079c89e)
-
To ensure MaxId is properly set for cell scalar array, use Insert instead of SetTuple. (cherry picked from commit a4fd5b72)
-
(cherry picked from commit 379a4c6d)
-
When the size of an array is fixed, there is no reason to use Allocate() since SetNumberOfTuples()/Values() is simpler and ensures MaxId is set. (cherry picked from commit 55d8f0ea)
-
This ensures that we aren't reading into values that are beyond the MaxId of the array. (cherry picked from commit 8aa534aa)
-
Calling SetNumberOfTuples() already does the calculations that were being done here in the code, and has the benefit of also adjusting the MaxId of the array. (cherry picked from commit e6fb41b7)
-
Added SetNumberOfValues() after the allocation in order to avoid setting values beyond the MaxId of the array. (cherry picked from commit 60fc8551)
-
The tuple, component indices into the weight array were reversed. (cherry picked from commit 7a12eaf8)
-
Setting the size of an array is best done with SetNumberOfValues(), since Resize() only changes the allocation but not the number of items present. (cherry picked from commit 6637fd6c)
-
This class was accessing its ItemLocation array beyond MaxId, now it uses the array in a more standard way. (cherry picked from commit 432053f7)
-
