Neither was needed as they were both empty and did nothing - the
compiler generated destructor can do this and it fixes the Clang++
warnings on the dashboards.

Change-Id: If5bbaa77669058c82f8ad6a66256a9b712aa753e

This wasn't needed, and is not implemented by any of the other classes
deriving from vtkTuple.

Change-Id: I00c36e7b6384cd14f6076e8cafcb16982dd136c3

64e7b2b9 COMP: Fix warnings caused by a backslash at the end of a comment.

Change-Id: Ibd9659d3bce4e6676b0feacf9d6acb0306b6e3da

29bbbaf2 DOC: Describe how to get L and U out of the LU factorization.

b47332bb Fixed reference counting bug in vtkTransform

c544585c Make "graphic error" api legacy

4d5cbfb6 OpenGL driver version detection and OS Mesa features

8e5b524b Shapefiles: If active layer specified, load only it.

* pvvtk/pv-master:

Change-Id: Ief9cc41049f1fdadcfc28951a81d54807ad319b3

Utkarsh Ayachit authored Jul 26, 2013 and Kitware Robot committed Jul 26, 2013

6e9be2bf Merge topic 'update_vtk_master' into master
f7f82536 Fixed regression introduced by a6fa3e37.
f81dd42f Merge topic '0012114-fix-getarrayindex-documentation' into master
7915505c Merge topic 'vtkFLUENTReader_array_indexing_12683' into master
0a2836c7 Fix off by one error in vtkFLUENTReader::GetCellsAscii().
736a4ade BUG:0012144 correct the documentation for GetArrayIndex
2193b802 Merge topic 'clear-std-isfinite' into master
f1760ede COMP: Clear VTK_HAS_STD_ISFINITE and related variables.
5239d41d Merge topic 'more-register' into master
c4b4eb3b Merge topic 'std-isfinite-detection' into master
0c41fe46 Merge topic 'coordinates_in_tooltip' into master
3e490c8c Removed deprecated and useless 'register' keyword
a9abb469 add row/column info to tooltip
f808e0c6 14164: Fix incorrectly set VTK_HAS_STD_ISFINITE variable.
c5f21d95 Merge topic 'update-kwsys' into master
3d6e5ff5 Merge branch 'upstream-kwsys' into update-kwsys
...

Utkarsh Ayachit authored Jul 26, 2013 and Kitware Robot committed Jul 26, 2013

e6bab494 Merge topic 'fix_dashboard_issues' into master
70ead432 Merge topic 'data-array-asserts' into master
14c8487e Safer fake coordinates for structured from unstructured CF grids
4b20f9b1 Increasing timeout for test.
46a51e1c Merge topic 'opengl-error-hunt' into master
6455c581 OpenGL error hunt -- dashboard cleanup
7e49ccd4 Merge topic 'opengl-error-hunt' into master
0994d6ca OpenGL error hunt -- dashboard cleanup
bc424e27 Add asserts to vtkDataArrayTemplate Set/GetValue()
e07e9824 Merge topic 'opengl-error-hunt' into master
62311d5e OpenGL error hunt -- dashboard cleanup
8be9b7a5 Merge topic 'opengl-error-hunt' into master
5e3e2043 Merge topic 'Fix_OctreePointLocator_Descr' into master
3545b515 Merge topic 'netcdf_units' into master
7cbcd88c fix windows compilation
5de3f32a Merge topic 'geojson_writer' into master
...

06ef3b9e Change `vtkScalarsToColors::SetAnnotations` behavior.

16543376 Cleanup logic for indexed colors.
37a10c60 Release graphics resources properly.
1509669d Cleaning up transfer function widgets.

Now SetAnnotations performs a deep copy of annotations
rather than referencing the provided array pointers.
This addresses ParaView bug 14189, where changes to the
array after SetAnnotations were not being reflected in
the map.

Change-Id: I8e78b30a5757eb9429dd5e02c9126cb5c01892c5

vtkDiscretizableColorTransferFunction used control points to map indexed colors.
To make applications simpler, added a new API to add indexed colors which are
used when present when in IndexedLookup mode. If no IndexedColors are specified,
the vtkDiscretizableColorTransferFunction reverts to the previous behaviour for
backwards compatibility.

Change-Id: If5b6791baccf2ee0022fd1b1098093d226ef64c3

vtkScalarBarActor wasn't releasing graphics resources for internal texture. This
fixes the problem.

Change-Id: I5f01c7d8ad0f78c25b4ab8f21bc8b4e66ae91c5f

Cleaning up issues with the context-items related to tranfer function editor.

+ Added support for log-mapped color-transfer function. We ensure that when
  using log-scale, the color transfer function widget also enter log-space.

+ vtkPiecewiseControlPointsItem had a bug resulting it deleting of end points
  even when told otherwise. Fixed that.

+ Added new event vtkControlPointsItem::PointsModifiedEvent which if fired when
  control points are changed making it easier for application to do something
  when the points are changed.

+ Improved the usability by adding ability to label control points.

Change-Id: If02bc1139f6ccb3d7c2198b07d10491f902bafda

There was a reference counting error in vtkTransformConcatenation when
a transform was DeepCopy'ed and shortened and then DeepCopy'ed and lengthened
using some of the same concatenated transforms.

When the list of concatenated transforms was shortened, the values
were not cleared out, letting pointers in the list that did not have a reference
according to the reference counting.  Later, when the list was set to a larger
list again and happenned to have the same transform at postion x, since there
was already a reference to the transform in the list, it did not increment
the reference count even though it should have.  This commit fixes this
by setting pointers beyond the end of the list to NULL when DeepCopy shrinks the list.

This commit also adds a regression test for that will segfault unless the
bug is fixed due to a reference count that is accidentally decremented in
a loop due to the bug. Running this test without this patch will segfault.

Change-Id: I27d7c2097f85a2f8c1d6e8a0e8213d041cc83125

Decorated vtkRenderWindow "graphic error" methods with legacy.
These were superceded by vtkOpenGL{Check/Clear}ErrorMacro's.

Change-Id: I2b3fbbbb1125b7611d4553e6ea1640b3f74121f6

Added support to vtkOpenGLExtensionManager for fine grained
OpenGL driver version detection. Added a flag to enable/disable
features with known driver specific bugs for testing new
driver releases. This mostly to test future Mesa releases to
determine if bugs have been fixed. Used the new feature to
see if some long standing Mesa bugs have been fixed in newer
Mesa releases for the OS Mesa renderer.

* NPOT texture support used by vtkOpenGLContextDevice2D and
vtkOpenGL2ContextDevice2D fixed in Mesa >= 8

* reset glReadBuffer to quiet incomplete FBO error report in
vtkOpenGLVolumeGPURayCastMapper

* vtkOpenGLVolumeGPURayCastMapper tests pass in Mesa >= 9
except TestGPURayCastCompositeShadeMask. Explicitly disabled
the test for OS Mesa.

* disable FBO for Mesa < 8 because of TestBlurAndSobelPasses
TestSobelGradientMagnitudePass test failures

* disable vtkOpenGLHAVSVolumeMapper for OS Mesa without
Gallium llvmpipe renderer

* vtkCompositeRGBAPass needs GL_ARB_texture_float. Mesa
may or may not be built with it enabled due to patent
restrictions and a bug in 8-9 releases.

* Disable FBO for Mesa drivers reporting OpenGL version 1.4
because of incomplete/buggy GL_ARB_draw_buffers.

* Although associated ctests run and pass, disable
vtkOpenGLVolumeTextureMapper3D for OS Mesa < 9 because of
false OpenGL invalid enum report.

Change-Id: Ic82aeb4747137aead53251026744baa61ced66d0

080a7f72 resolve new warning

This commit eliminates the following warning:
  warning C4018: '<' : signed/unsigned mismatch

Change-Id: I3dc5f506e5b08ea6365d71487c5b39364faabec3

2c86574b use SetMaxDiscreteValues for categorical color map
02ffb7ac change MAX_DISCRETE_VALUES into an ivar

f7f82536 Fixed regression introduced by a6fa3e37.

a6fa3e37 changed the logic in vtkWin32OpenGLRenderWindow::CleanUpRenderers().
That method was used when swithcing from offscreen to onscreen. That resulted in
the renderers not having correct references to render window after switching
from offscreen to onscreen context causing dashboard failures on ParaView
dashboards. This fixes that issue.

Since the a6fa3e37 changed the logic in CreateOffScreenDC() by not using
CleanUpRenderers() to reset renderers, it makes sence to change the logic in
ResumeScreenRendering() to not use CleanUpRenderers() for resetting renderers as
well.

Change-Id: I32e58854fa8720c99212cee5d7a2652f11dc9b21

Use the new function vtkAbstractArray::SetMaxDiscreteValues within
vtkTreeHeatmapItem.  This allows us to properly color categorical
datasets that contain more than 32 distinct values.

Within this commit, we also change the color set that we're using for
the categorical color map.  The accent set is pretty, but it only has
8 different values.  Instead we'll use qualitative set3 because it
has 12 values.

Finally, we fix a bug in vtkTreeHeatmapItem so that our NaN color
(grey) is used appropriately by our categorical lookup table.

Change-Id: I53bc9f13240be8c0c10b5dbc77bdcca1ea0fa2a1

Change MAX_DISCRETE_VALUES from a hardcoded value of 32 to a user-
settable ivar.  This allows users of vtkAbstractArray to generate
a categorical color map for data sets containing more than 32
discrete values.

For backwards compatibility, we keep the MAX_DISCRETE_VALUES enum
and use it to initialize MaxDiscreteValues.  This enum is marked
as deprecated.

Change-Id: I83bf7c598288dc28a4ad231778c66e1571e3a461

736a4ade BUG:0012144 correct the documentation for GetArrayIndex

0a2836c7 Fix off by one error in vtkFLUENTReader::GetCellsAscii().

Thanks to Dan Clark to pointing out the issue in the vtkFLUENTReader.

Change-Id: I0bea0c4ed74917d6af913d95f8cfd9a6d45dec11

Per bug # 0012114, the documentation was wrong.

Change-Id: I4386cc01edc6f84f9ca1aa2a02d19ff61d1d4bdf

f1760ede COMP: Clear VTK_HAS_STD_ISFINITE and related variables.

The cmake variables VTK_HAS_STD_ISFINITE, VTK_HAS_STD_ISNAN, and
VTK_HAS_STD_ISINF were set incorrectly prior to commit c4b4eb3b,
and the incorrect settings are stuck in the CMakeCache.txt of many
machines (and will be until they clear their cache).  So this commit
specifically removes those variables from the cache, in order to
allow the correct checks to be done.  This commit will be reverted
after one week.

Change-Id: I27073fbe199957a551c65c090313d74a850b28db

3e490c8c Removed deprecated and useless 'register' keyword

f808e0c6 14164: Fix incorrectly set VTK_HAS_STD_ISFINITE variable.

a9abb469 add row/column info to tooltip

Fixed clang dashboard warning.

Change-Id: I89056ee91d0aade22bd129367fe3b8fbaf9ab8b5

The tooltip displayed while hovering over a heatmap cell now
includes the name of the row & column.  This makes it easier for
a user of this class to tell what data value they're looking at
when they are zoomed in on part of the item.

Change-Id: Idd3df2bcd98986d49ddaeee0f2b9353209c7f67f

This change is necessary for compilation with gcc in C++11 mode.

The check_symbol_exists() macro that we had been using to find
std::isfinite, std::isnan, and std::isinf always failed, so it
wasn't doing anything useful.  Both this macro and the newer
check_cxx_symbol_exists() are unsuitable for std::isfinite etc.
because these function might exist either as function templates,
or as simple function overloads, depending on the compiler.

As a side note, the existing checks for the c99 functions, e.g.
  cmake_symbol_exists(isnan "math.h" VTK_HAS_ISNAN)
should instead use this:
  cmake_cxx_symbol_exists(isnan "cmath" VTK_HAS_ISNAN)
because vtkMath.h includes cmath, not math.h, and on some systems
math.h provides these function while cmath does not (or, cmath
might only provide them within the std namespace).  In fact, most
instances of cmake_symbol_exists() in VTK should be replaced by
cmake_cxx_symbol_exists() when VTK adopts a more recent CMake.

Change-Id: I1ef7cd4daa52067badb0928d01c3f8b22953c601