Release 2.0

The 2.0 release brings a major new feature, support for integers, as well as extremely significant performance increase when it comes to handling of loops.

Currently the only type of integers supported is u8. The user has a choice of three overflow semantics.

The performance increase is both theoretical and practical. The worst-case time complexity has been improved from O((2^N)*((2^N)^3)) to O((2^N)^3). Furthermore, whereas the average time complexity in the previous release is the same as the worst-case time complexity, in the new release the cubic factor is closer to linear for typical programs.

Precompiled binaries are available for AArch64 macOS and x86_64 Linux; these are the only two machines I have right now.

Initial Public Release (v1.0)

This is the initial public release of the inference tool.

Precompiled binaries are available for the x86_64 architecture on macOS, Linux, and Windows. Download them below.

(For these obsessed with small binaries: it's totally possible to make the binaries smaller, but I haven't bothered.)