Asymmetric numeral systems: entropy encoding (2013)

This was popularized by Yann Collet's work on https://github.com/Cyan4973/FiniteStateEntropy

Almost all current implementations (except TurboANX) are in general not better than huffman coding. See: https://sites.google.com/site/powturbo/entropy-coder

Better test this yourself on your own data with TurboBench: https://github.com/powturbo/TurboBench

detailed list of implementations found http://encode.ru/threads/2078-List-of-Asymmetric-Numeral-Sys...

Does anyone know if this is patent-encumbered?

(2013). Already in practical use.

Does it get us the advantages of arithmetic coding without the patents?

Is arithmetic coding still patented? I remember hearing about it being patented along with GIF back in like the '90s, but patents don't last that long - it's 2016, even GIF is out of patent now.

Man, that was more complex than I thought. Looks like various methods of doing the coding were patented at different times. https://en.wikipedia.org/wiki/Arithmetic_coding#US_patents All the ones listed there are expired, so that's a good sign I guess.

Woohoo!