David E. DeMarle authored 6 years ago and Kitware Robot committed 6 years ago

597cc33a fix ios build issues
c4751b8f Increment version to VTK 8.1.1
8fa3de4e comparison was being done for the wrong case
3588d82c Remove vtkProbeOpebGLSupport executable from Mobile
61694e1f Check that context exists before trying to pop context.
3538ec08 Use public instead of protected member functions
373b2a55 fixed case where the array name differed from
98a23692 fix issues with the array calc and string arrays
...

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !4284

The framework was using the vtk version but it was not set
yet in the CMakeLists file.

Old 32 bit architecture was on by default. Rmeoved.

(cherry picked from commit bc4923bd)

It only needs to be checked when the IdType IS 64bit
not when it is 32bit

(cherry picked from commit f0d637bf)

Remove it as it's target is for desktop and it avoids
link challenges.

(cherry picked from commit 2d6d5772)

On some linux drivers (such as nvidia version 384.111 and 387.34) setting the
context to zero causes a segfault so check before setting in cases where the
destruction may be trying to pop to a zero context.

(cherry picked from commit a7988f5c)

(cherry picked from commit 29f8f9ff)

the variable name. Shoudl fix dsashboard issues.

(cherry picked from commit 7279cebc)

The old code did not correctly handle DataObjects
that had string arrays in them. It would cause
the index to be off after the string array. So it
might look liek it worked but the data would
be incorrect. Very bad. This fixes PV issue
paraview/paraview#12967

(cherry picked from commit 81f882c4)

Added an executable that is always built regardless
of if testing is on that checkes for OpenGL2 support.
This can be used to help diagnose problems when users
have a build using VTK that is not working and testing
is off as it is by default.

(cherry picked from commit 9eba3269)

Newer OSX versions no longer support it

(cherry picked from commit f8a9eba1)

Fix issue where if someone set relative coincident params on
a dataaset mapper they had no effect.

(cherry picked from commit a565bf8e)

Fix an issue with the JPEGReader where it could not
handle images larger than 2GB and also consumed more
memory than it should.

Fix issue in vtkOpenGLTexture when resizing textures larger
than 2GB.

(cherry picked from commit 79436154)

for android superbuilds

(cherry picked from commit 01a5b8e5)

Path from gitlab Constantine @Butakoff

(cherry picked from commit 8fb6f1d5)

Add a header file function to handle some of the work in
testing opengl support and selecting mesa. Also add a cmake option
to turn on delayed loading for opengl.

(cherry picked from commit 29607014)

When handling a 3d volume it was accessing too far when
there was cell data present.

(cherry picked from commit 428d603a)

David E. DeMarle authored 6 years ago and Kitware Robot committed 6 years ago

fd6b906b 17265: Fix OpenGL2 vtkImageResliceMapper OOB issue

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !4076

David E. DeMarle authored 6 years ago and Kitware Robot committed 6 years ago

20c61dd6 Backport fix for issue #17211 to release.

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !4241

David E. DeMarle authored 6 years ago and Kitware Robot committed 6 years ago

53f83b43 Clamp tolerance to range [0.0, VTK_DOUBLE_MAX]
21e0e07c Fix logic when tolerance is 0

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !4264

David E. DeMarle authored 6 years ago and Kitware Robot committed 6 years ago

508c2ada Backport 'reslicemapper-border' to release branch.

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !4275

When MultiSamples are used in VTK (which is the default on Windows
and Linux), a thin border appeared at the edge of the image if the
the image pixel values were not zero at the edge.  The thin border
is the result of the zero 'Background' value of vtkImageReslice.
All of the "Background" pixels produced by vtkImageReslice are
supposed to map to beyond edges of the polygon used to render the
image texture.  However, MultiSamples causes the edge to be soft,
and the coloring of these background pixels bleeds into the edge
of the polygon.

The solution is to tell vtkImageReslice to add a bit more edge
tolerance when clamping its interpolation coordinates.  This moves
the 'Background' beyond the edge of the polygon, so that these
background pixels no longer bleed into view.

(cherry picked from commit fa9b8ddee39be035a27e47a9211ca4ff7cc278ba)

If t == 1.0, the logic for determining that t is 1.0 was incorrect
when the tolerance is 0.0.

(cherry picked from commit 21cffcb58ec27d0c707d0038ad0d6b9ccef4089a)

Andreas Buykx authored 7 years ago and David Gobbi committed 6 years ago

The vtkCellPicker::IntersectActorWithLine() method called on
composite data could generate invalid minCellId and minSubId.

David E. DeMarle authored 6 years ago and Kitware Robot committed 6 years ago

4d586107 Retain renderer to recompute right rendered rectangle.

Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !4232

The billboard bounds were broken by bad bookkeeping.

Revised to retain the relevant renderer to recalculate the right
rectangle right away without really rendering.

Cory Quammen authored 7 years ago and Kitware Robot committed 7 years ago

77183b27 Fix out-of-bounds memory write

Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: David E. DeMarle <dave.demarle@kitware.com>
Acked-by: Sean McBride <sean@rogue-research.com>
Merge-request: !4129

An out-of-bounds memory write could occur in GetSingleLoop() when the
condition (pointCells->GetNumberOfIds() < 2). GetSingleLoop() adds a
line cell to complete a loop in this case, but that fact was not
accounted for in the caller when allocating the argument for the
lineBool array. Fixed that, and cleaned up the code by using a
std::vector instead of an array. Finally, added a test that exercises
the condition to make sure it is tested.

For OpenGL2, the vtkOpenGLImageSliceMapper:RenderPolygon was being
called even if the slice specified by vtkImageResliceMapper was
beyond the volume bounds.  This caused a regression: specifically,
when the slice moved out-of-bounds the old polygon would be reused
even though no polygon should be displayed at all.  The fix is for
RenderPolygon to return early if no polygon should be displayed.

Closes #17265

Sean McBride authored 7 years ago and Kitware Robot committed 7 years ago

e6f0a9b1 Fixed a clang Wfloat-conversion warning in a public header

Acked-by: Kitware Robot <kwrobot@kitware.com>
Reviewed-by: Allison Vacanti <allison.vacanti@kitware.com>
Merge-request: !4042

Sean McBride authored 7 years ago and Kitware Robot committed 7 years ago

24c108de change tag for release branch libTIFF import
cac41974 Merge branch 'upstream-tiff' into release-tiff-mangle-fixes-via-script
7f55ffcd tiff 2018-02-14 (7ae879be)

Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Ben Boeckel <ben.boeckel@kitware.com>
Merge-request: !4046

Sean McBride authored 7 years ago and Kitware Robot committed 7 years ago

9acc86b8 Fix clear shader replacement issue
1e45e062 add missing modifieds when changing shader replacements

Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Ben Boeckel <ben.boeckel@kitware.com>
Merge-request: !4040

* upstream-tiff:
  tiff 2018-02-14 (7ae879be)

Kitware Robot authored 7 years ago and Sean McBride committed 7 years ago

Code extracted from:

    https://gitlab.kitware.com/third-party/tiff.git

at commit 7ae879be23194d845859246f261926a4ddc5c572 (for/vtk-release).

Of all the pubilc VTK headers my own app includes, this was the only such warning.  Fixing it allows me to more easily keep the warning on for my own code.

Sankhesh Jhaveri authored 7 years ago and Sean McBride committed 7 years ago

The replacement would never be erased because of a faulty if check.
Added test code to the relevant tests for regression testing in the
future.

(cherry picked from commit f4f12b96)

Ken Martin authored 7 years ago and Sean McBride committed 7 years ago

Modifieds were missing causing shaders to not be rebuilt
when they should have been

(cherry picked from commit d592cb2a)

6b255c3b KWSys: Backport build interface include directory fix to VTK 8.1

Merge-request: !3955