azrafe7 / hxorderedmap

Ordered maps for haxe 4.0.0-preview5+.

hxOrderedMap

Ordered maps for haxe 4.0.0-preview5+.

Overview

This is a multi-type abstract that wraps haxe.ds.Map and can be used with the same API.

It also adds a few methods/props (so far length, keysCopy(), clear()).

The main difference with Map is that the order of insertion is preserved.

Minimal example

var keys = ["1", "apple", "banana", "orange", "one", "two", "three"];
var values = [for (i in 0...keys.length) i];

var omap = new OrderedMap();
var map = new Map();

for (i in 0...keys.length) {
  var k = keys[i];
  var v = values[i];

  omap[k] = v;
  map[k] = v;
}

trace("ORDERED  : " + omap); // always  [1 => 0, apple => 1, banana => 2, orange => 3, one => 4, two => 5, three => 6]
trace("UNORDERED: " + map);  // may be  {orange => 3, three => 6, apple => 1, banana => 2, one => 4, 1 => 0, two => 5}

Notes

Internal book-keeping to maintain the order of insertion makes using this data structure a bit slower than using a Map directly (e.g. approximately 1.5-2x slower for set/remove).

Retrieving some info from an OrderedMap instance is faster for the same reason (e.g. getting the number of key-values, a copy of the keys, clearing the map, etc.).

Benchmarks for latest build (https://azrafe7.github.io/haxe-bencharts/?repo=azrafe7/hxOrderedMap).

Note that this is not intended as a replacement for haxe.ds.Map (especially if you care about speed), but I've found it useful in many situations.

License (MIT)

See LICENSE.

hxorderedmap's Issues

Find how to properly cast/unify OrderedMap to Map (so they can be used interchangeably)

Somewhat not-too-minimal example of what I'd like to do here: https://gist.github.com/azrafe7/d6f22eff5faae82af9c3ee34cc5be1d1

For now I'm trying to work around it by using a getInnerMap() method.

(Note-to-self: basing the work-around on this messy thing https://gist.github.com/azrafe7/059f56b27c869331c31b7208adea399a)

[cpp] Extending IMap interface doesn't work as I would expect

UPDATE(02 nov 2018): I have a work-around in place that makes the lib work on cpp too. Still not totally happy with it and trying to reduce the case so I can submit a proper issue upstream.

ORIGINAL POST(31 oct 2018):
This most probably is my bad somewhere, although I haven't found a way to solve it yet.
Other targets are ok with it.

Currently I'm just including the IMap interface into the IOrderedMap one (for cpp).