randn.dev

What It Is

RandN is a random number generation library for .NET, positioned as a modern replacement for .NET Random. The documentation states that it aims to address common pitfalls in Random around performance, statistical quality, and API design. It also draws inspiration from Rust’s Rand crate, and can be understood as a kind of “NodaTime for random numbers.”

Core Capabilities

Functionally, RandN provides several categories of RNGs: general-purpose StandardRng, SmallRng, and ThreadLocalRng, as well as reproducible implementations such as ChaCha, PCG32, Mersenne Twister, and XorShift. StandardRng and ThreadLocalRng are marked as cryptographically secure, while SmallRng is designed for higher speed and a smaller state. For scenarios that require fixed-seed reproducibility in experiments or tests, the documentation explains that types implementing IReproducibleRngFactory guarantee the same sequence from the same seed.

Distributions, API, and Ecosystem

RandN is more than a generator for random bytes or integers. It also provides distributions such as Bernoulli, Uniform, Unit Interval, and Singleton, with support for integers, floating-point types, decimal, TimeSpan, and more. The API examples are clear: for instance, Bernoulli.FromRatio(8, 10) represents an 80% probability, while Uniform.NewInclusive(1, 6) represents a die roll. It is installed via NuGet and offers GitHub resources, .NET Fiddle examples, and RandomShim, making it easier to gradually migrate existing code that depends on Random.

Pricing and Documentation

The source material does not mention commercial pricing; RandN is presented as a GitHub and NuGet library. Whether it has a specific open-source license is not made clear in the captured content. Overall documentation quality is good, organized by versions from v0.1.x to v0.5.x, and covers Usage, RNGs, Distributions, Reproducibility, and Extending RandN. It also flags deprecated components, such as recommending SystemCryptoRng instead of CryptoServiceProvider.

Pros, Cons, and Who It’s For

Its strengths are broad algorithm coverage, an API design that is less error-prone than using Random directly, and support for cryptographic security, performance, and reproducibility needs. Its drawbacks are that the source text lacks performance benchmarks, license details, information about the maintenance team, and support channels. Some algorithms are also noted as potentially moving out of the main package in the future. RandN is suitable for .NET backends, games, simulations, test data generation, probabilistic sampling, and secure random number generation scenarios.

Access from China

The source text does not describe access from China. NuGet and GitHub may be affected by network instability in mainland China, but this alone is not enough to determine the project site’s availability. If access is limited, alternatives include .NET’s built-in RandomNumberGenerator, Random, or other NuGet libraries for random numbers and statistical distributions.