GitLab

Change-Id: I80d4dbda38c7a6d83d6ba3fadeb81cce1154a1b1

Change-Id: I7e5c55259159752882c8ae056e08f9bec12b3c02

The XML readers and writers, both parallel and serial, now work
with the pipeline changes. Now, all writers work with pieces
even though the structured ones can handle asking for a sub-
extent.

Change-Id: I9f757ba12a3b9fb84e141506dacadad608eb8b99

Refactoring the way VTK goes between piece and structured
extents. Before, extent translators were used when the pipeline
moved from structured to unstructured data converting piece
request to extent request. This caused many problems with filters
that altered extents, mainly a lot of redundant IO due to
repartitioning of different extents. This became extremely
cumbersome to manage when running distributed. The new behavior
pushes the extent translation all to way to the readers and
only when readers are able to read a subset. This works much
better. The only downside is that filters need to be able to
handle data extents different than update extents. Most filters
can do this but many imaging filters cannot. Those that are
needed in parallel will have to be updated.

As part of this work, I also removed MAXIMUM_NUMBER_OF_PIECES
and added CAN_HANDLE_PIECE_REQUEST. MAXIMUM_NUMBER_OF_PIECES had
reduced to being a boolean. 1 for serial sources, -1 for parallel
sources. I removed it and added a CAN_HANDLE_PIECE_REQUEST instead.
This key, produced by a source, tells the executive that a source is
able to handle piece request. It is a source only key produced in
RequestInformation and is not propagated downstream. If this key is
not present, the executive will only execute the source for piece 0
to produce the entire data. It is then up to the user to add a filter
that splits the data for other piece requests. The only exception to
this is when CAN_PRODUCE_SUB_EXTENT is present, in which case the
executive will split using an extent translator AT THE source - not
downstream.

Conflicts:
	Filters/FlowPaths/vtkParticleTracerBase.cxx

Change-Id: I8db4040289ff87331adeecded4a738313d9b52df

Jacob Becker authored May 16, 2014 and David Gobbi committed May 16, 2014

When only one polyline was left, it would erroneously not be pushed
onto the "incomplete" list.

Also, a vector normalization had no check for division-by-zero, causing
a NaN if the vector length was zero.

Change-Id: I36a9b7a3561e6a490c795806e1b35c10de44a446

Change-Id: Idf4f870e2046dafe3c5bf8aa2ca9c5f2f6d79a37

The ForwardTransformPoint method was missing a "!" in a null pointer
check, causing the transform to perform an identity transform.  The
tests for this class did not cover this code path, even though it is
the most common path in the general usage of this class.  A new test
has been added, and an old test has been re-enabled.

Also, add the missing GetDisplacementScale method.

Change-Id: I3cc21fda11e4c4f0a83dc020e0e9968086246d45

This commit should resolve any "R is already initialized" errors.

Change-Id: Ie1814e1dffaa1a975b5305df0a1a66d3edff5304

Change-Id: I99c13233c9870f25b34947e071a9a23e33d710c2

The use of atexit() was causing R code that uses the parallel
library to fail nondeterministically.

Change-Id: Iff8eaafd786341f7bc10d50338c50525be492385

Change-Id: I9c90b018f0f9477462b28f970b30d15084005ce4

It is unused otherwise.

Change-Id: Ic13a1e15ae2dd121b0122e7b520551184da57e2c

Change-Id: Ic1ae3928ee171531136d42630fec6484eb0a9cbd

If one table and one tree were requested as output from
vtkRCalculatorFilter, its output would be a vtkTable or vtkTree.
It actually needs to be a vtkMultiPieceDataSet to contain both
objects though, and this would result in a segmentation fault
shortly thereafter.  This commit fixes this degenerate case
and adds a test for it.

Change-Id: I53e468bba2e2d336dc7a4a7f9fb53fdcb64f98b6

vtkPStreamTracer was causing the creation of a new locator at
every iteration (task), which is very slow for cell locators
since they have to bin data every time. Fixed by removing
the reallocation and using one locator. The input data does not
change between iterations so recreationg the locator was totally
unnecessary.

Change-Id: I966f25caf2e05ba08355a426f3e31f1ac66517ef

The PrintSelf method of vtkStructuredImplicitConnectivity
was causing failures in the otherPrint test b/c it was
dereferencing NULL ivars. Fixed that by adding guards
to chekc if the ivars are NULL.

Change-Id: I50416653e14d5af50a6408c09129f1deebf63ebf

Modified the function names in vtkStructuredData to use "points"
instead of "nodes" in order to be consistent with the rest of VTK.
Further, updated all affected code that is using vtkStructuredData
to reflect this change.

Change-Id: Iad0a876b49e0b5439e47b963c73885c2ade3c0d6

Updated tests in the vtkFiltersParallelMPI module to reflect changes in
VTK's test macros.

Change-Id: I81d3983a35f3fe2e49cb2b83f2771d2c92084ff2

This commit introduces the vtkStructuredImplicitConnectivity class, intended
as a lower-level helper that provides functionality for resolving
situations where "gaps" are present in a distributed structured dataset. Gaps
are resolved by communicating the data (growing the grid to the right) as
necessary to close the gap. This situation is common with finite difference
grids that are partitioned with respect to the nodes of the grid, or when some
other post-processing filter, e.g., sampling the grid to reduce the resolution
introduces a gap.The present implementation requires one grid per process and
can deal with 2-D datasets -- XY,XZ,YZ -- as well as 3-D datasets. A test for
all 2-D and 3-D cases is also added.

Change-Id: I5260ce4fb5ced45434a719c8aacc2153253917a7

In commit cd4fc15a (Rewrite the testing macros, 2014-03-14)
the vtk_test_mpi_executable() helper was introduced to add
MPI tests.  It depends on the vtk_mpi_link() helper in the
vtkMPI module.  Include the latter in all test CMakeLists.txt
files that add MPI tests.

Change-Id: Ifecc7c6b1bf856a8f66905bfb2d69e22b5de3aac

The TEST_DEPENDS module option already takes care of this.

Change-Id: Ib27e0d7a6d362f9358c512b0f6fb934cc35605a4

indentation was messed up in a block.

Change-Id: I7b1890aa552f8b4133f07cca515f858a809deea9

Before this commit, vtkRCalculatorFilter produced the same type
of output as its input.  This is no longer the case.  For example,
you can now run an R script that takes one table as input
and produces two trees as output.  A test is included for this
new functionality.

This commit also makes vtkRAdapter::RToVTKTree() more robust.
Previously, we assumed that the elements in the R Tree would
appear in a certain order, but apparently that is not guaranteed
to be true.  We now parse these elements in whatever order they
appear.  While I was modifying this function, I also cleaned it up
so it should now adhere to VTK's coding style.

Change-Id: I988361ee8e6788f6f51e0c930fc52df50d1b2912

Change-Id: I851265b552c48e752906f0ff4e109c3a6e6fff9c

Change-Id: I45908373e619b6cce723f4411280687bdb0caf87

Variable was only used in an assert() and so caused
an unused variable warning in release.

Change-Id: I58e9cba411ae55a3b8a1a28d1b5ddc2d173a6a9d

Managing separate files with the new functions is really messy.

Change-Id: Id9cbc2a0de08a4ddc8760d1f924cbbf5af1af3df

Change-Id: I6546ff164baca86afb1a73b896dd10b1bce2c005

Change-Id: I5f19b101063e1a25d4f5db5c4356ce5efffbd90d

Change-Id: If92879d67d11dfc5965ef42a8c173539801b3c98

Change-Id: I10a54acbd076e055d8071414f2fff1ebafe1653d

Fix typos, style & a potential memory leaks.
Add initial support for extent on level 0 rectilinear grid.

Change-Id: I03ddd5a0b5a30901054a59ebe2bee2be38fa2bf8

When converting from an R data.frame to a vtkTable, store their
row names (if any) in a string column at the beginning of the table.

Change-Id: Ie9bc73a8c83ae7aec46ce05bc0de849a1c0f3da2

Fix the way how the bags are computed. They are now performed at
50 and 99% of the distribution.
Functional bag plots also create a median serie which is the
median line of all input series. Outliers series are also
tagged with an "_outlier" extension in their name.

Change-Id: I990428c9b79278be9be725491232b3e9d69190b0

This commit adds the DuplicateNodes property to the structured
RCB partitioners, which controls where the nodes between
partition interfaces will be duplicated or not. By default the
nodes are duplicated, which yields partitions that are abbuting
at the interfaces sharing a face. If the DuplicateNodes property
is switched off, there will be a "gap" between partitions, which
is common with Finite Difference codes in particular.

Change-Id: If57529e60f608444dbd055d62d1b1d863e22b924

vtkDataSetSurfaceFilter now ignore output cells that contain all
ghost points. Note that this is a partial implementation and only
works for unstructured grids and 3D cells.

Change-Id: If335a9ecf5e9db1700695118fcec6e77bfe58720

Since vtkTable does not have a concept of row names, we add a
vtkStringArray containing the names as the first column of the
vtkTable.

Change-Id: I2c9e20ed2c1d9cc21205e137ce9b9cfedd02a816

Use precomputed table of densities to avoid precision differences
on some machines.
Also remove unuseful normalization step in the plot and algorithm.

Change-Id: Ifb28b82b276ee93b66f4ddf8f4375f7b3b465b94

Will Schroeder authored Mar 03, 2014 and Ben Boeckel committed Mar 10, 2014

Change-Id: I7fb57f3a45922d9106e7395329d62e5f060b1417

Previously the filter just passed points through. This behavior is an
option now (PassPointsOn) and instead unused points (after extraction) are
discarded. This implementation is consistent with vtkExtractGeometry and
is the new default behavior (PassPointsOff).

Change-Id: I8fb396caa236e54c0afdb69f94a0ea785e8bca97