RFR: JDK-8302800: Augment NaN handling tests of FDLIBM methods

[Joe Darcy]

On Mon, 20 Feb 2023 10:22:48 GMT, Raffaello Giulietti <rgiulietti at openjdk.org> wrote:

>> Augment NaN-handling tests to include exotic NaN bit patterns. Assuming this changeset goes back first, I'll update https://github.com/openjdk/jdk/pull/12608 to follow the same convention afterward.
>
> test/jdk/java/lang/Math/Tests.java line 41:
> 
>> 39:     private Tests(){}; // do not instantiate
>> 40: 
>> 41:     static volatile double zero = 0.0;
> 
> Not directly related to this PR and just out of curiosity: is there a reason for this to be `volatile` and non-`final`?

The comment on the PLATFORM_NAN field was mean to address that -- an attempt to make sure the expression to create the platform NaN gets evaluated at runtime rather than being constant-folded at compile time.

-------------

PR: https://git.openjdk.org/jdk/pull/12644