Rtree

=================

- find_library and libspatialindex library loading :PR:`131`

=================

- Refactored tests to be based on unittest :PR:`129`
- Update libspatialindex library loading code to adapt previous behavior :PR:`128`
- Empty data streams throw exceptions and do not partially construct indexes :PR:`127`

=================

- Add Index.GetResultSetOffset()
- Add Index.contains() method for object and id (requires libspatialindex 1.9.3+) :PR:`116`
- Add Index.Flush() :PR:`107`
- Add TPRTree index support (thanks :user:`sdhiscocks` :PR:`117`)
- Return container sizes without returning objects :PR:`90`
- Add set_result_limit and set_result_offset for Index paging :commit:`44ad21aecd3f7b49314b9be12f3334d8bae7e827`

Bug fixes:

- Better exceptions in cases where stream functions throw :PR:`80`
- Migrated CI platform to Azure Pipelines https://dev.azure.com/hobuinc/rtree/_build?definitionId=5
- Minor test enhancements and fixups. Both libspatialindex 1.8.5 and libspatialindex 1.9.3 are tested with CI

===============

- Support for Python 3 added.

=================

- 0.7.0 relies on libspatialindex 1.7.1+.
- int64_t's should be used for IDs instead of uint64_t (requires libspatialindex 1.7.1 C API changes)
- Fix __version__
- More documentation at http://toblerity.github.com/rtree/
- Class documentation at http://toblerity.github.com/rtree/class.html
- Tweaks for PyPy compatibility. Still not compatible yet, however.
- Custom storage support by Mattias (requires libspatialindex 1.7.1)

=================

- 0.6.0 relies on libspatialindex 1.5.0+.
- :py:meth:`~rtree.index.Index.intersection` and :py:meth:`~rtree.index.Index.nearest` methods return iterators over results instead of
lists.
- Number of results for :py:meth:`~rtree.index.Index.nearest` defaults to 1.
- libsidx C library of 0.5.0 removed and included in libspatialindex
- objects="raw" in :py:meth:`~rtree.index.Index.intersection` to return the object sent in (for speed).
- :py:meth:`~rtree.index.Index.count` method to return the intersection count without the overhead
of returning a list (thanks Leonard Norrgård).
- Improved bulk loading performance
- Supposedly no memory leaks :)
- Many other performance tweaks (see docs).
- Bulk loader supports interleaved coordinates
- Leaf queries. You can return the box and ids of the leaf nodes of the index.
Useful for visualization, etc.
- Many more docstrings, sphinx docs, etc