Mashi - A compressor for JS/WASM 64k intros

 Sagacity writes at Pouet.net:

<<Hi friends, I just released the first version of Mashi, which is a compressor for JS/WASM 64k intros. It is based on the core compression routines of Squishy by Ferris/Logicoma, which I'm sure most of you are familiar with.

It is intended as a replacement for Compeko or JsExe or any other kind of tool that relies on the browser's built-in Deflate implementation. I built it primarily for use for compression of WASM (with a tiny amount of JS needed for browser interop) and for that reason it adds a full WASM context model that really improves compression of those kinds of binaries. The decompressor is hand-written in WASM so if you're interested to see what hell looks like, check out "decompress.wat" in the repo linked below.

However, intros that are pure JS can also benefit, although in my unscientific testing the results varied. 0b5vr's 0mix intro went from 64k to 48k, but Luminosity by Bypass was almost unchanged. So YMMV.

Anyway, feel free to test it out. Binaries are available at https://github.com/datatrash/mashi/releases and the source code is at https://github.com/datatrash/mashi

It's all a bit untested (especially under Linux and on non-Chrome-based browsers), but I thought that with Revision coming our way it could already be of use to someone.

Btw, pull requests that reduce final binary size are of course very welcome :)

And a big shout out to Ferris for allowing me to use Squishy as a basis for this, obviously!>> 

Comments

Post a Comment

Popular posts from this blog

Volko Hull - An Alternative to Convex Hull

When you have a point cloud and want to obtain the shape, you can use the convex hull or the alpha shape algorithm. In the computational geometry library I am using in my projects, CGAL, these algorithms are built-in, but they have a disadvantage: the points are not output in the correct order. I first tried to solve this problem by implementing a convex hull algorithm myself that outputs the points in the correct order, but this did not always help since it ignored concave corners. Therefore I wrote my own algorithm to replace it. My algorithm divides the rectangular area of the given cluster into four triangles and computes the points belonging to the outer border of each triangle. Thus it also includes concave corners. Imagine this to be the rectangle: -------- |\    /| | \ A/ | | C\/  | |  /\D | | /B \ | |/    \| -------- If it is a square, we can assume that its sides have a length of 2a. The coordinates of the mid point are always (0, 0). Th...
Read more

JavaScript: How to serialize maps of maps of maps of ...

Recently I had to serialize a map in JavaScript. On the Internet I found code similar to this: /* json-map.js */ export const replacer = (key, value) => {   if(value instanceof Map) {     return {       dataType: 'Map',       value: Array.from(value.entries()),     };   } else {     return value;   } } export const reviver = (key, value) => {   if(typeof value === 'object' && value !== null) {     if (value.dataType === 'Map') {       return new Map(value.value);     }   }   return value; }   Now I had the problem that this works for maps of arrays, but not for maps of maps. I pondered over it for a while and then came up with a recursive solution that does not only work for maps of maps, but also for maps of maps of maps, maps of maps of maps of maps, and so on: /* json-map.js */ export const replacer = (key, value) => {   if(value inst...
Read more