Skip to content

std::map/std::set implementation using the adaptive radix tree

License

Notifications You must be signed in to change notification settings

justinasvd/art_map

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

65 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Introduction

This library is a C++14 template library and, as such, there is no library to build and install. It is heavily based on Laurynas Biveinis' excellent implementation of an adaptive radix tree [1]. See his implementation of in-memory database at https://github.com/laurynas-biveinis/unodb.

There are notable differences from the original implementation. They mostly stem from the fact that both art::set and art::map are aimed to be drop-in replacements for std::set and std::map respectively. This means that art::set and art::map must meet the requirements of Container, AllocatorAwareContainer, AssociativeContainer and ReversibleContainer, which unodb does not try to do.

Caveats

There are limits to what kinds of keys can be used in ART. Keys must be transformable to BitwiseComparable, maintaining the order imposed by std::less<Key> or std::greater<Key>. Such transformations for integral key types and pointers are already provided. In theory, it is possible to provide bitwise orderings for floating point types and, perhaps, composite types as well. We have not ventured there. Patches welcome.

Also, insertions and deletions on an Adaptive Radix Tree can cause nodes to resize. The result is that insertions and deletions can invalidate iterators pointing to values other than the one being inserted/deleted. This is notably different from libstdc++ set/map which take care to not invalidate iterators on insert/erase except, of course, for iterators pointing to the value being erased.

Requirements

The code uses SSE4.1 intrinsics (Nehalem and higher). This is in contrast to the original ART paper needing SSE2 only.

The following minimum versions are required to use the library:

  • GCC 6.1
  • Clang 3.4

Also, the code depends on Boost libraries. Minimum Boost version 1.60 is recommended.

See Building tests and benchmarks.

Building tests and benchmarks

This describes the build process using cmake. As pre-requisites, you'll need git and cmake installed.

# Check out the library.
$ git clone https://github.com/justinasvd/art_map.git
# Go to the library root directory
$ cd art_map
# Make a build directory to place the build output.
$ cmake -E make_directory "build"
# Generate build system files with cmake.
$ cmake -E chdir "build" cmake -DCMAKE_BUILD_TYPE=Release ../
# or, starting with CMake 3.13, use a simpler form:
# cmake -DCMAKE_BUILD_TYPE=Release -S . -B "build"
# Build tests and benchmarks.
$ cmake --build "build" --config Release

References

[1] Leis, Viktor, Alfons Kemper, and Thomas Neumann. "The adaptive radix tree: ARTful indexing for main-memory databases." 2013 IEEE 29th International Conference on Data Engineering (ICDE). IEEE, 2013.

About

std::map/std::set implementation using the adaptive radix tree

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published