https://github.com/brianhicks/elm-avl-exploration

AVL trees for Set and Dict in Elm - optimized for reads
https://github.com/brianhicks/elm-avl-exploration

Last synced: 6 days ago
JSON representation

AVL trees for Set and Dict in Elm - optimized for reads

• Host: GitHub
• URL: https://github.com/brianhicks/elm-avl-exploration
• Owner: BrianHicks
• License: bsd-3-clause
• Created: 2017-02-13T15:49:06.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2017-02-24T21:00:16.000Z (almost 8 years ago)
• Last Synced: 2024-12-09T19:53:30.318Z (14 days ago)
• Language: Elm
• Homepage: http://package.elm-lang.org/packages/BrianHicks/elm-avl-exploration/latest
• Size: 193 KB
• Stars: 1
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Elm AVL Exploration [![Build Status](https://travis-ci.org/BrianHicks/elm-avl-exploration.svg?branch=master)](https://travis-ci.org/BrianHicks/elm-avl-exploration)

A drop-in reimplementation of Elm's [`Dict`](http://package.elm-lang.org/packages/elm-lang/core/latest/Dict) using [AVL trees](https://en.wikipedia.org/wiki/AVL_tree) instead of [red-black trees](https://en.wikipedia.org/wiki/Red%E2%80%93black_tree).

## Quick Start

1. `elm package install BrianHicks/elm-avl-exploration`

2. `import Dict.Avl as Dict` wherever you want to use AVL Dicts

## Why?

I wrote the original implementation [for a blog series on functional data structures](https://www.brianthicks.com/post/2016/11/13/functional-sets-part-1-construction/).

After I was done, I wondered how my implementation compared to core, speed-wise.

Short answer: it's about 30% faster for reads, but 30% slower for inserts.

If you know that you do a lot more reading than you do writing, especially on a small number of keys, this library will probably be faster than core.

Data from my machine is below. You can replicate by building `benchmark/Main.elm` and visiting the result in your browser.

**WARNING: Treat these figures as hypothetical.**

You can and should use [`BrianHicks/elm-benchmark`](https://github.com/BrianHicks/elm-benchmark) to verify these results in your own project with your own sample data.

### `get`

`Dict.Avl.get` is about 30% faster than `Dict.get`, on average.

![get performance](docs/get-performance.png)

### `toList`

`Dict.Avl.toList` is about 10% faster than `Dict.toList`, on average.

`toList` is implemented using `foldl`, which may be an indicator of performance there.

This advantage is mainly found on sets of 10 or fewer items.

![toList performance](docs/toList-performance.png)

### `insert`

`Dict.Avl.insert` is about 35% slower than `Dict.insert`, on average.

This disadvantage is mainly found on sets of 10 or more items.

![insert performance](docs/insert-performance.png)

### `remove`

`Dict.Avl.remove` is about 25% faster than `Dict.insert`, on average.

This advantage is mainly found on sets of 10 or fewer items.

![remove performance](docs/remove-performance.png)

## License

Elm AVL exploration is licensed under a 3-Clause BSD License, located at [LICENSE](LICENSE).