Refactor picking to use one thread per entity

Added triangle visitor by refactor the triangle volume extractor
Added ability to do a ray intersection with a single volume without using a thread
Added entity collector
Use one thread for each entity
Ignore entities that don’t have a pick object
Removed list of triangle volumes stored on each entity
Use triangle visitor to test each triangle by creating a single volume on the stack
Added triangle and 3 vertex index to the pick event

Change-Id: Id3fd53549bf3ea07805426ef868df174f30e176b
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>

1 files changed, 2 insertions, 0 deletions

diff --git a/src/render/jobs/calcboundingvolumejob.cpp b/src/render/jobs/calcboundingvolumejob.cpp
index feeb7c422..ccffe2c84 100644
--- a/src/render/jobs/calcboundingvolumejob.cpp
+++ b/src/render/jobs/calcboundingvolumejob.cpp
@@ -123,9 +123,11 @@ void calculateLocalBoundingVolume(NodeManagers *manager, Entity *node)
                     const int stride = pickVolumeAttribute->byteStride() ? pickVolumeAttribute->byteStride() : sizeof(float) * pickVolumeAttribute->dataSize();
                     QVector<QVector3D> vertices(pickVolumeAttribute->count());
 
+                    // TODO avoid copying the vertices
                     for (int c = 0, vC = vertices.size(); c < vC; ++c) {
                         QVector3D v;
                         const float *fptr = reinterpret_cast<const float*>(rawBuffer);
+                        // TODO unwrap loop (switch?)
                         for (uint i = 0, m = qMin(pickVolumeAttribute->dataSize(), 3U); i < m; ++i)
                             v[i] = fptr[i];
                         vertices[c] = v;