Support wide lines in 3.2 using a geometry shader. This patch
includes a number of cleanups to make using geometry shaders
easier etc.

Add an option to Ready a shader program based on a vector of shaders

Rename some functions to match what they do

Cleanup some memory leaks in the LIC code

vtkgl::CellBO renamed vtkOpenGLHelper
vtkgl::substitute moved to vtkShaderProgram::Substitute
Creation of vtkOpenGLIndexBufferObject
Creation of vtkOpenGLVertexBufferObject
Move methods that were in vtkglVBOHelper into new classes
Move IndexCount from CellBO to vtkOpenGLIndexBufferObject
etc

As we get closer to a next VTK release, make sure this is
fairly clean or at least cleaner than it was.

This should fix kamino by alowing the software renderer

Should fix compiler warnings on dashlin1

With buildbot's default timeout at 180, these tests time out on some
machines.

Also removed some dead code as a consequence,
like setting the deallocated pointer to null
then to a new address.

Also removed setting some ivars to null in dtors.

Fixed all clang -Wimplicit-fallthrough warnings:
- Some fixed by adding missing breaks
- Other fixed by adding VTK_FALLTHROUGH statement

Under (modern) clang with C++11:
 #define VTK_FALLTHROUGH [[clang::fallthrough]]

Otherwise:
 #define VTK_FALLTHROUGH ((void)0)

VTK_FALLTHROUGH serves two purposes:
 - under clang, suppresses -Wimplicit-fallthrough warnings
 - makes code more self-documenting

Change-Id: Ib635b7c83f7b061dcc47a28463ce4a326e4860c1

One fix is to properly report errors when a shader program fails to
link.  The other is to fix rendering molecules when using light kit
or positional lights.

Change-Id: I74d938f2986f421c61ffdcf03e69a6b728ffe484

The old code had some mathematically issues that were
exposed on a few dashbaords.  This should fix it.

Change-Id: Icd5f8e6296a23d9d336e29fccdd85f13082f9d85

Change-Id: I3caa31d8c121a47ed49065609f0aa12c30a6c7a4

This class is a faster implementation of vtkCompositePolyDataMapper
that requires its inputs to adhear to some constraints. The big change
is that it required adding functions to append to a VBO or IBO.

Part of this change is I decoupled the opacity from scalar coloring.
Previously a change in opacity required rebuilding the VBO if
scalar coloring was on. While this is still the behavior, it can
now be optimized to not do that.  It doesn't made sense for
the property's opacity to be mixed in with scalar colors, it
was just done that way because that we all the old pipeline
could do.

Also cleaned up a few long lines while I was a it.

Change-Id: I3a06b502f154c4da7465e53eeeaeab9492bca88f

A few changes here. The motivation is to support using the MCDCMatrix
in the vertex shader where possible. This has the advantage of
performing the MCVC x VCDC multiplication in double precision
on the CPU as opposed to float precision on the shader. This fixes
a couple test failures.

Along with that change cleanup a bit how we determine what uniforms
to set. There were some uniforms being set when they did not exist
in the program. Provided a more general approach for passing that
type of information between the code that builds the shader program
and the code that sets the uniforms.

Add a substitute helper function that does the same job as replace
but returns a bool if a replacement was made. Also operates on
the source string by reference.

Cleaned up some long lines to be shorter.

Change-Id: Ic23133679f1d40d51c29f11810d29baf85edd306

This change breaks the Update VBO method into a method
to GetNeedToRebuildBufferObjects and BuildBufferObjects to make
it easier for subclasses to override the default behavior.

Change-Id: I99b3378b0e6b7c3ec9da4000ab33d213b8fc67c4

Used the following command:

find . \( -iname *.h -and -type f \) -exec perl -pi -w -e 's/__vtk(.*)_h/vtk$1_h/g;' {} \;

Fixes many, but not all, clang -Wreserved-id-macro warnings.

Change-Id: I694ff053cb204c70e37b261296f496064d91fc07

Used the following command:

find . \( -iname *.in -and -type f \) -exec perl -pi -w -e 's/__vtk(.*)_h/vtk$1_h/g;' {} \;

Fixes many, but not all, clang -Wreserved-id-macro warnings.

Change-Id: I86db91ad530c203347f2b0dfd29c31f843c3f2b7

Add in a mapper that renders points as gaussian splats.  In
the future we can extend it a bit more to do other useful
things. It colors the points and scales then based on
arrays that are passed in.

This patch also includes a performance improvement for
the spheremapper so that it uses less memory.  It is
the same basic code as the point gaussian mapper which
is why it is in the same patch.

Change-Id: I8bce4446be063eb5d6a5eb4bc1a51e4369ba61cf

Change-Id: I6f150f184244e2dc2c348759793960b6b47f81f8

A few fixes and cleanups. Use the Blinn-Phong specular lighting
model in the LightKit and Positional shaders. Old OpenGL uses this
model and it is a decent model. Remind myself that the camera is
positioned at 0,0,0 in view coordinates, had to fix a couple places
that were making different assumptions.

Minor changes to TestVBOPLYMapper to make it easier to use
when testing out changes.

Minor warning and documentation fixes as well.

Change-Id: I8bf108ec18fef2f0a2f83ada8cdf7663086036be

Imposters use FragDepth as opposed to FragCoord.z

Change-Id: If376c2cb3c0dc8f1513b2712690d773afea4027a

Change-Id: I23740eea54093433b1b4bf67ce9d2f38dd6e9164

Somefragment shaders want to modify vertexVC which previously
was a hard coded varying that is used by the lighting calcs.
Made it no longer hard coded. Fixes imposters for Intel chips.

Change-Id: I3feae834a0b6c9a7eb775404b083bdaba55dfb49

Add in a new directory ChemistryOpenGL that includes a factory
override for vtkMoleculeMapper that uses imposters to draw the
spheres and cylinders. Supports picking.

Change-Id: I718cc4fbae25c3dc7ab82bf19ed7a12e472c51b2

Change-Id: I76848acf2f0939f63df6c3194b302eebd6a68009

Change-Id: I48bdaf653e677ca3626ed6087bca25831fed1b26

Change-Id: I9f2bafb8707c86180a0893280474c101593cc9c0

Lots of places where we pass floating point values as integers; make the
loss of precision more explicit.

Change-Id: Icafde395f0463421b5f36d1e4e855fa25cab5ab4

These are great for some basic benchmarks, but cause timeouts on some
Gerrit hosts using Mesa at least. Commenting out for now.

Change-Id: Ia68fc6c3c26b367bcd3d04ab7ac93b0ff050b56b

This test is taking too long with OpenGL backends, and failing with
time outs. Reduced the number of renders, and simply use vtkNew for
the timer that was added.

Change-Id: I43d1a969346f6f9d4032032b34cd62437c81def0

This adds a concept of a rendering backend to the module API, with
an OpenGL and OpenGL2 backend. They are mutually exclusive and one
or the other should be selected (or None in the case when rendering
is not requested). This needs a little more work as the default
backend enables rendering when the group is disabled at this point.

Change-Id: I98d660e063ef58d52698e08ab06dd45601bea9cd

Change-Id: I834e94a505fdb209dc29793931a9c68b0e676111

Change-Id: Ie6e63c47b03e2aeb8ee1137e0ca333c98b86f651

Change-Id: I10a54acbd076e055d8071414f2fff1ebafe1653d