algebr4

Vectors and Matrices for up to 4 dimensions written in typescript.

Builds

The packaged npm version provides different builds for different module systems:

The default build is for ES6 modules which can be used for angular projects among others:

import { Vec3 } from '@b3gm/algebr4';

const v = new Vec3(3.0, 4.0, 5.0);

If you want to use the library in the backend for a regular node project, you will have to import the commonjs build:

import { Vec3 } from '@b3gm/algebr4/dist/commonjs/algebr4';

const v = new Vec3(3.0, 4.0, 5.0);

And finally there is a bundled build for direct use in the browser, that creates the top level algebr4 namespace within the global namespace in dist/bundle/algebr4.bundle.js.

<script src="https://cdn.jsdelivr.net/npm/@b3gm/[email protected]/dist/bundle/algebr4.bundle.js"></script>
<script type="text/javascript">
  var v = new algebr4.Vec3(3.0, 4.0, 5.0);
</script>

General Naming Conventions

Typescript interfaces use the following prefixes:

• L - Literal: base interfaces for all other interfaces and classes. All methods take Literals as input. Those interfaces contain no methods and only declare the used fields. This facilitates assignment and usage of object literals or JSON deserialized objects as input for all vector and matrix methods.
• R - Readable: extend Literals. Contain no methods, that change the internal state of the object. All methods create and return new readable objects. Note that this does not mean Immutable, since the object might still be writable.
• W - Writable: extend Readables. All methods from the Readable interface should be overwritten to return writables instead of readables. Most methods from the readable interface have "Self" suffixed overloads, that work on the object itself and update their internal state instead of returning a new object.

Unsafe Methods

Some methods have overloads, that are suffixed with Unsafe. They only work correctly if their inputs satisfy certain criteria, but no such checks are undertaken upon executing those. For vectors and quaternions, this means, they have to have a length of 1. Rotation matrices need to have a determinante of +-1 and need to be orthogonal.

Using unsafe methods with parameters, that do not satisfy those criteria, will therefore yield unexpected results, but can safe performance if for instance a vector's length is already guaranteed to be 1.0 by surrounding code.