Merge branch 'master' of https://github.com/rust-lang-nursery/rand

Author: Rohit Joshi

CHANGELOG.md

@@ -9,6 +9,19 @@ A [separate changelog is kept for rand_core](rand_core/CHANGELOG.md).
 You may also find the [Update Guide](UPDATING.md) useful.
 
 
+## [0.5.1] - Unreleased
+
+### Platform support and `OsRng`
+- Remove blanket Unix implementation. (#484)
+- Remove Wasm unimplemented stub. (#484)
+- Dragonfly BSD: read from `/dev/random`. (#484)
+- Bitrig: use `getentropy` like OpenBSD. (#484)
+- Solaris: (untested) use `getrandom` if available, otherwise `/dev/random`. (#484)
+- Emscripten, `stdweb`: split the read up in chunks. (#484)
+- Emscripten, Haiku: don't do an extra blocking read from `/dev/random`. (#484)
+- Linux, NetBSD, Solaris: read in blocking mode on first use in `fill_bytes`. (#484)
+- Fuchsia, CloudABI: fix compilation (broken in Rand 0.5). (#484)
+
 ## [0.5.0] - 2018-05-21
 
 ### Crate features and organisation

README.md

@@ -108,15 +108,16 @@ Rand release if required, but the change must be noted in the changelog.
 
 ## Crate Features
 
-Rand is built with only the `std` feature anabled by default. The following
+Rand is built with only the `std` feature enabled by default. The following
 optional features are available:
 
 - `alloc` can be used instead of `std` to provide `Vec` and `Box`.
 - `i128_support` enables support for generating `u128` and `i128` values.
 - `log` enables some logging via the `log` crate.
 - `nightly` enables all unstable features (`i128_support`).
 - `serde1` enables serialization for some types, via Serde version 1.
-- `stdweb` enables support for `OsRng` on WASM via stdweb.
+- `stdweb` enables support for `OsRng` on `wasm-unknown-unknown` via `stdweb`
+  combined with `cargo-web`.
 
 `no_std` mode is activated by setting `default-features = false`; this removes
 functionality depending on `std`:

src/distributions/cauchy.rs

@@ -50,12 +50,9 @@ impl Cauchy {
 impl Distribution<f64> for Cauchy {
     fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> f64 {
         // sample from [0, 1)
-        let mut x = rng.gen::<f64>();
-        // guard against the extremely unlikely case we get the invalid 0.5
-        while x == 0.5 {
-            x = rng.gen::<f64>();
-        }
+        let x = rng.gen::<f64>();
         // get standard cauchy random number
+        // note that π/2 is not exactly representable, even if x=0.5 the result is finite
         let comp_dev = (PI * x).tan();
         // shift and scale according to parameters
         let result = self.median + self.scale * comp_dev;