- BUG FIX: Fix middleSplit_ for same points by @yzabalotski in #250
- Fix build warnings.
- Add conan install instructions.
- Add multiple thread kdtree build support for KDTreeEigenMatrixAdaptor (PR #246)
- BUG FIX: nanoflann::SearchParameters::sorted was ignored for RadiusResultSet.
- ResultSet classes now must implement a sort() method.
- Added type IndexType to nanoflann:KDTreeBaseClass
- Potentially more efficient scheduling of multi-thread index building (PR #236)
- Bump minimum required cmake version to 3.5 (PR #230)
- Fix outdated NANOFLANN_VERSION macro in header file
- Fix poll-allocator alignment problems
- Add NANOFLANN_USE_SYSTEM_GTEST option
- Look for Threads dependency in CMake config script
- Other changes:
- Save one redundant call to
computeMinMax()
inmiddleSplit_
(PR#220 by qq422216549). This saves a lot of time, up to 20% faster in a benchmark with small (thousands) point clouds.
- Save one redundant call to
- API changes:
- Add new search method
rknnSearch()
for knn searches with a maximum radius. - Add missing
SearchParameters
argument toKDTreeSingleIndexDynamicAdaptor_::knnSearch()
(PR#213 by ManosPapadakis95). - Add missing method
KNNResultSet::empty()
for consistency with the other result sets.
- Add new search method
- Other changes:
- Add GUI examples for each search type:
nanoflann_gui_example_R3_knn
nanoflann_gui_example_R3_radius
nanoflann_gui_example_R3_rknn
- Add GUI examples for each search type:
- API changes:
- Users of radius search should change their result placeholder type:
std::vector<std::pair<IndexType, DistanceType>>
=>std::vector<nanoflann::ResultItem<IndexType, DistanceType>>
. (See #166 for the motivation of this change). - More concise auxiliary (internal) type name:
array_or_vector_selector
->array_or_vector
. - Remove obsolete parameter
nChecks_IGNORED
. Removed fromSearchParams
constructor too, so that structure has been renamedSearchParameters
to enforce users to update the code and avoid mistakes with the order of its ctor parameters. - Added method RadiusResultSet::empty()
- Template argument rename:
AccesorType
=>IndexType
(does not actually affect user code at all). - Added concurrent tree building support, refer to
KDTreeSingleIndexAdaptorParams::n_thread_build
.
- Users of radius search should change their result placeholder type:
- Other changes:
- Macros to avoid conflicts with X11 symbols.
- Inline an auxiliary example function in case users want to use it and include the file in multiple translation units (Closes #182).
- Move all benchmarking code, data, and scripts to its own repository to keep this repo as clean as possible.
- Fix "potentially uninitialized" GCC warning.
- Clarified, even more, in docs and examples, that L2 distances are squared distances.
- Removed the (with modern compilers) now useless
inline
keyword in class members. - Add examples with GUI (requires mrpt-gui):
- nanoflann_gui_example_R3: Radius search on R³ Euclidean space.
- nanoflann_gui_example_bearings: NN search on non-Euclidean spaces.
- BUGFIXES:
- Avoid segfault if saving an empty index (Closes #205).
- Added flag SkipInitialBuildIndex to allow not wasting time building a tree when it will be loaded from a file later on (PR #171).
- Mark all constructors explicit, to avoid unintended creation of temporary objects (Issue #179).
- BUGFIX: avoid potential index out of bounds in KDTreeSingleIndexDynamicAdaptor (PR #173)
- Install pkg-config .pc file under lib directory (Closes #161).
- Integrate AppVeyor CI.
- Fix incorrect install directory for cmake target & config files.
- Do not install example binaries with
make install
. - Provide working examples for cmake and pkgconfig under
examples/example_*
directories.
- nanoflann::KDTreeSingleIndexAdaptor() ctor now forwards additional parameters to the metric class, enabling custom dynamic metrics.
- Add and apply a
.clang-format
file (same one than used in MOLAorg/MOLA projects). - Examples: clean up and code modernization.
- CMake variables prefixed now with
NANOFLANN_
for easier integration of nanoflann as a Git submodule. - Fixes for IndexType which are not of integral types PR #154
- save/load API upgraded from C
FILE*
to C++ file streams (By Dominic Kempf, Heidelberg University, PR).
- Add optional argument for Eigen matrix layout commit.
- Throw exception on malloc failure PR #126.
- Respect GNUInstallDirs in CMake install rules PR #131.
- Fixed bug in KDTreeSingleIndexDynamicAdaptor. See: https://github.com/jlblancoc/nanoflann/commit/a066148517d16c173954dcde13c1527481b9fad3
- Fix build in XCode.
- Simplify CMakeLists for Eigen example (requires Eigen3Config.cmake now)
- Avoid setting cmake global executable build path
- Instructions for
make install
for Linux and Windows (Closes #87). - Fix all (?) MSVC conversion warnings (Closes: #95).
- Avoid need for _USE_MATH_DEFINES in MSVC (Closes: #96)
- Eigen::Matrix datasets: now uses std::cref() to store a reference to matrix.
- GSOC2017 contributions by Pranjal Kumar Rai:
- Support for dynamic datasets.
- Support for non-Euclidean spaces: SO(2), SO(3)
- Fixed: split plane now correctly chooses the dimensions with the largest span. Should lead to more optimal trees.
- knnSearch() now also returns the number of valid points found.
- Fix potential compiler warnings if
IndexType
is signed. - New unit tests comparing the results to those of brute force search.
- Fixed: many classes constructors get const ref arguments but stored const values.
- Added KDTreeSingleIndexAdaptor::radiusSearchCustomCallback() (Based on a suggestion by Yannick Morin-Rivest)
- Better documentation in class headers.
- Cleanup of unused code.
- Parameter KDTreeSingleIndexAdaptorParams::dim has been removed since it was redundant.
- Created hidden constructors in nanoflann class, to disallow unintentional copies which will corrupt the internal pointers.
- Fixed crash if trying to build an index of an empty dataset.
- Two internal containers are now automatically defined as fixed-size arrays if the problem dimension is known at compile time, improving efficiency. The new/modified datatypes are: KDTreeSingleIndexAdaptor::BoundingBox, KDTreeSingleIndexAdaptor::distance_vector_t
- Fixed compilation with GCC 4.8 and C++11 enabled (Thanks to Simon Praetorius).
- Fixed warnings about unused parameters.
- Fixed L1_adaptor.accum_dist(), which implemented L2 instead (Closes #1)
- Fixed wrong typedef in KDTreeEigenMatrixAdaptor<> for IndexType!=size_t (Closes: #2)
- Fixed: Memory pool wasn't freed after each call to buildIndex()
- GCC: Added -isystem flag to gtest headers to avoid pedantic warnings.
- Fixed compilation with Visual Studio 11 (MSVC 2012).
- Fixed compilation of gtest with VS11 and its _VARIADIC_MAX "bug".
- Added a security check to launch an exception if searches are attempted before buildIndex().
- New example to demonstrate save/load the index to files.
- save/load methods exposed as public.
- GTest sources are now embedded, due to the changes in newer Ubuntu packages which don't carry the precompiled libs.
- Added asserts to detect whether the user passes NULL as query points.
- New method RadiusResultSet::worst_item()
- New method RadiusResultSet::set_radius_and_clear()
- Avoid potential collision of min/max macros with <windows.h>
- Removed unneeded #include's of std headers.
- New sample code for vectors of vectors.
- Fixed building of tests for MSVC in Windows.
- Allow manually setting the path to Eigen3 (mainly for building examples under Windows).
- Better documentation and added graphs of a benchmarking for helping choosing "leaf_max_size".
- Now KDTreeSingleIndexAdaptor::buildIndex() can be called several times even when the dataset size changes (Thanks to Rob McDonald for reporting!)
- Some fixes to kd_tree index and L1/L2 metrics to allow distinct types in data elements and in the distances. This is mainly to permit elements being vectors of integers (e.g. uint8_t) but distances being real numbers.
- Examples and unit tests have been corrected to use template arguments instead of being hard-wired to "float" data types (Thanks Thomas Vincent for noticing!).
- Fixed warnings for MSVC and for GCC with "-Wall -pedantic"
- Updated performance tests to work with the final nanoflann code (they were written for a very early version).
- All main classes now have new template arguments for the type of indice, which now defaults to "size_t" instead of "int". In case this breaks backward compatibility in user code, especify "int" to override the default template arguments, although "size_t" it's recommended.
- Initial version