Merge pull request #219 from Alexhuszagh/issue_218

Patch use of `set_precision` on i586.

.github/workflows/Cross.yml

@@ -24,6 +24,8 @@ jobs:
           - aarch64-unknown-linux-gnu
           - arm-unknown-linux-gnueabi
           - armv7-unknown-linux-gnueabihf
+          # NOTE: Required for issue #218
+          - i586-unknown-linux-gnu
           - i686-unknown-linux-gnu
           - i686-unknown-linux-musl
           # NOTE: This fails on cross v0.2.1, which is unusual since

CHANGELOG

@@ -31,6 +31,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Patch doctests with TODOs in dependencies (#222).
 - CI with OSS-Fuzz for the deprecated `actions/upload-artifact@v3` (#221).
+- Float parsing on `i586` targets (#219).
 - Bug where the `radix` feature wasn't enabling `power-of-two` in `lexical-core` or `lexical` (#204).
 - Fixed performance issues due to a lack of inlining on the Eisel-Lemire algorithm (#210).
 - Issue with parsing non-decimal exponent radixes when using a decimal mantissa radix for floating-point numbers (#212).

lexical-parse-float/src/number.rs

@@ -74,7 +74,7 @@ impl Number<'_> {
         // function takes care of setting the precision on architectures which
         // require setting it by changing the global state (like the control word of the
         // x87 FPU).
-        let _cw: () = set_precision::<F>();
+        let _cw = set_precision::<F>();
 
         if self.is_fast_path::<F, FORMAT>() {
             let radix = format.radix();