On Wed, 6 Mar 2024 20:49:55 GMT, Joe Darcy <[email protected]> wrote:
>> The atan2 and hypot cases added would fail for a particular test library
>> that has errors in the millions of ulps for those inputs, rather than the
>> small number of ulps specified for the error.
>
> Joe Darcy has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Remove unneeded lines.
Looks good
test/jdk/java/lang/Math/WorstCaseTests.java line 622:
> 620: int failures = 0;
> 621: // Cannot represent exact result, allow 1 additional ulp on top
> of documented bound, rounding up.
> 622: double ulps = 3.0;
Suggestion:
double ulps = 3.0; // 1.5 + 1.0, rounded up
-------------
PR Review: https://git.openjdk.org/jdk/pull/18140#pullrequestreview-1927836589
PR Review Comment: https://git.openjdk.org/jdk/pull/18140#discussion_r1519723927
