Skip to content

An arbitrary-precision numerics library, ported from LLVM to Zig with a C API.

License

Notifications You must be signed in to change notification settings

vezel-dev/libap

libap

Warning

This is currently in-development vaporware.

libap is a port of LLVM's arbitrary-precision (AP) numerics types to Zig, with an additional C API on top to allow C programs to make use of the library as well.

Note that the types offered by this library have a fixed, predetermined bit width and semantics; they do not grow to maintain precision. For example, when creating an ap.Int, you have to specify exactly how many bits it should occupy. Any operation that produces a new value will have the same bit width, and any operation involving two ap.Int values requires that they have the same bit width. Contrast with the types in std.math.big which grow as necessary.

Most applications that use arbitrary-precision numbers do want them to grow in order to maintain full precision. In other words, most users should look to std.math.big (or an equivalent C library). The types offered by libap are primarily useful for compilers, emulators, cryptography, and other fields where the desired precision is bounded and/or strict semantics are required.

Status

libap currently covers the following LLVM types:

Changes made to these types in LLVM are periodically synchronized here. The current master code is based on LLVM commit bbc6504b3d2f237ed7e84dcaecb228bf2124f72e.

Usage

The minimum Zig version supported by this project can be found in the minimum_zig_version field of the build.zig.zon file. We generally try to track the latest release of Zig.

To use libap in your Zig project, first add it as a package:

zig fetch --save=ap https://github.com/vezel-dev/libap/archive/vX.Y.Z.tar.gz
# Or, to use Git:
zig fetch --save=ap git+https://github.com/vezel-dev/libap.git#vX.Y.Z

(You can find the latest version on the releases page.)

Then, consume the ap module in your build.zig:

const ap = b.dependency("ap", .{
    .target = target,
    .optimize = optimize,
});
exe.root_module.addImport("ap", ap.module("ap"));

You should now be able to @import("ap") in your Zig code.

Installation

If you plan to use libap as a C library, you will likely want to install it somewhere on your system since C does not have a package manager. To do so, grab a source code archive from the releases page, and run zig build --prefix $HOME/.local -Doptimize=ReleaseFast or similar.

The result will look like this:

$ tree $HOME/.local
/home/alexrp/.local
├── include
│   └── ap
│       └── ap.h
└── lib
    ├── libap.a
    ├── libap.so
    └── pkgconfig
        └── libap.pc

Assuming you have set your C_INCLUDE_PATH, LD_LIBRARY_PATH, and PKG_CONFIG_PATH environment variables appropriately, you should now be able to use libap like any other C library.

License

This project is licensed under the terms found in LICENSE-LLVM, i.e. the same license as the C++ code from which this library was ported. This is the new LLVM license following the LLVM relicensing effort.

About

An arbitrary-precision numerics library, ported from LLVM to Zig with a C API.

Topics

Resources

License

Code of conduct

Security policy

Stars

Watchers

Forks

Releases

No releases published