Mathieu Westphal (Kitware) authored 2 weeks ago and Kitware Robot committed 2 weeks ago

4321af04 vtkOBJImporter: Fix bug in multi-file processing

Acked-by: Kitware Robot <kwrobot@kitware.com>
Tested-by: buildbot <buildbot@kitware.com>
Reviewed-by: Mathieu Westphal (Kitware) <mathieu.westphal@kitware.com>
Merge-request: !11954

Static variable was used to vertices indexing.
It causes incorrect start index for other than first file.

Vicente Bolea authored 2 weeks ago and Kitware Robot committed 2 weeks ago

dd88415c ci: remove duplicated MAVEN_NATIVE_ARTIFACTS

Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: buildbot <buildbot@kitware.com>
Reviewed-by: Sebastien Jourdain <sebastien.jourdain@kitware.com>
Merge-request: !11950

Dan Lipsa authored 2 weeks ago and Kitware Robot committed 2 weeks ago

0946a23a Avoid memory new on SetNumberOfTuples and delete on SetVoidArray

Acked-by: Kitware Robot <kwrobot@kitware.com>
Tested-by: buildbot <buildbot@kitware.com>
Merge-request: !11640

Dan Lipsa authored 2 weeks ago and Kitware Robot committed 2 weeks ago

6660255b Replace and with &&
a758f446 Do not load empty datasets
3df51351 CanReadFile for nc, grib, h5 and .zgroup (file for zarr folder)
2d477d62 Deal with 0 dimensions
29d5db76 Add missing type

Acked-by: Kitware Robot <kwrobot@kitware.com>
Tested-by: buildbot <buildbot@kitware.com>
Merge-request: !11948

The reason of this change being done in a different commit since this
change cannot be merged to master due to a conflict in an unrelated but
adjacent line at configure_options.cmake.

(cherry picked from commit 7a350f5b)

Ben Boeckel authored 2 weeks ago and Kitware Robot committed 2 weeks ago

aa2e0f9f vtkOpenXRSceneObserver: add missing `<algorithm>` header
ce7dc622 ci: exclude test failures on Linux with aarch64
a95f131b UnitTestMath: fully initialize `quat`
f8cf906b UnitTestMath: consider that `char` might be unsigned
700649df vtkHDFUtilities: support `char` as unsigned
ad3c8e73 IO/HDF: link tests to VTK::IOExodus for information key
46f0c203 TestHyperTreeGridSourceDistributed: use `int8_t` instead of `char`
46869d7b ci: exclude some precision-sensitive tests on aarch64
...

Acked-by: Kitware Robot <kwrobot@kitware.com>
Tested-by: buildbot <buildbot@kitware.com>
Reviewed-by: Cory Quammen <cory.quammen@kitware.com>
Acked-by: Johnny <johnnync13@gmail.com>
Merge-request: !11895

Mathieu Westphal (Kitware) authored 2 weeks ago and Kitware Robot committed 2 weeks ago

2d2993db Fix RGBA color scalar array name in vtkPLYReader

Acked-by: Kitware Robot <kwrobot@kitware.com>
Tested-by: buildbot <buildbot@kitware.com>
Reviewed-by: Mathieu Westphal (Kitware) <mathieu.westphal@kitware.com>
Merge-request: !11944

`char` is unsigned on aarch64.

`char` on aarch64 is `unsigned` and the negative variants for
`PointClassification` end up leaking out when compared with literals
where promotion occurs (as `int`). While `CellClassification` lacks
negative variants, using its enum name helps to keep usage consistent.

Assign the `int8_t` storage size for these `enum` types and store its
type explicitly where used.

GCC on Linux aarch64 is warning that the temporary doesn't live long
enough. I don't see why it wouldn't though as `.Get()` returns a pointer
which shouldn't need any lifetime extension rule to keep alive.

This code was rewritten in dda65158 (BUG: Fix escape character
handling, improve test coverage., 2010-11-24). Given the `- '0'` on the
value guarded by the conditionals, it is likely that these were meant to
be the ASCII value of the numerals 0-9, not the numeral values 0-9. Use
`isdigit` to simplify the expression.

Discovered via Linux aarch64 warnings about `*pos >= 0` being a
tautology.

On Linux aarch64, `char` is unsigned, so comparison with negative
integers is always false.

On Linux aarch64, `char` is unsigned, so integer promotion interferes
with using a literal `-1`.

This predicate clause was introduced in e6d1e71e (BUG: Avoiding debug
assertion on with VS8…, 2009-05-06) in the context of avoiding a debug
assertion in VS8 when using the routine during `CanReadFile` as it opens
in ASCII mode but might actually be a binary file.

On aarch64, this way of testing the high bit is a tautology. Instead,
use the `isascii` function to make the expression meaningful at a
glance.

This warning is about GCC7+ having a different ABI than GCC6. Since GCC6
is no longer supported, we can ignore this ABI difference.

This version has prebuilt binaries for Linux aarch64 available.

For use with aarch64 because the Fedora 39 container has issues
rebuilding due to the removal of the openturns repository.

Disable OSPRay and Anari on aarch64 as they use Intel-specific bits
currently.