|author||Jeff Vander Stoep <email@example.com>||Thu Nov 05 15:27:27 2020 +0100|
|committer||Jeff Vander Stoep <firstname.lastname@example.org>||Thu Nov 05 15:27:27 2020 +0100|
TEST_MAPPING: test dependers of this crate Bug: 168167373 Test: TH Change-Id: I0c297e9d0cbfb65399c3bc2d41b2834c2be1859e
Core traits and error types of the rand library, plus tools for implementing RNGs.
This crate is intended for use when implementing the core trait,
RngCore; it defines the core traits to be implemented as well as several small functions to aid in their implementation and types required for error handling.
The main rand crate re-exports most items defined in this crate, along with tools to convert the integer samples generated by
RngCore to many different applications (including sampling from restricted ranges, conversion to floating point, list permutations and secure initialisation of RNGs). Most users should prefer to use the main rand crate.
rand_core crate provides:
The traits and error types are also available via
The current version is:
rand_core = "0.5.0"
Rand libs have inter-dependencies and make use of the semver trick in order to make traits compatible across crate versions. (This is especially important for
SeedableRng.) A few crate releases are thus compatibility shims, depending on the next lib version (e.g.
0.3.1). This means, for example, that
rand_core_0_3_0::SeedableRng are distinct, incompatible traits, which can cause build errors. Usually, running
cargo update is enough to fix any issues.
alloc-only configurations, as well as full
std functionality. The differences between
no_std and full
std are small, comprising
RngCore support for
Box<R> types where
std::io::Read support for types supporting
RngCore, and extensions to the
Error type's functionality.
std feature is not enabled by default. This is primarily to avoid build problems where one crate implicitly requires
std support and another crate requires
std support. However, the
rand crate continues to enable
std support by default, both for itself and
serde1 feature can be used to derive
Deserialize for RNG implementations that use the
rand_core is distributed under the terms of both the MIT license and the Apache License (Version 2.0).