RFR CSR JDK-8202555: Double.toString(double) sometimes produces incorrect results

Brian Burkhalter brian.burkhalter at oracle.com
Tue Mar 17 21:13:04 UTC 2020

Hi Raffaello,

> On Mar 17, 2020, at 12:38 PM, Raffaello Giulietti <raffaello.giulietti at gmail.com> wrote:
> there's a new version of the documentation [1] for the attached patch to solve the long standing issues described in [2]. The CSR referred to in the subject is in [3].
> Besides many improvements in the proofs, a couple of simplifications and a better overall organization (this is my hope), I could find an ideal treatment of a special case that in the previous inception had a ugly, yet pragmatic solution in form of rather ad-hoc switch constructs. The new approach is mathematically clean and makes the code some lines shorter.

Thanks for cleaning up the background paper. I’ll plan to review initially at least the through the part I read before.

> The new code reflects this improvement and also adds a couple of additional tests.
> Passes tier1 tests.

A tier 1-3 test run is in progress.

> @BrianBurkalter
> Brian, can you please upload the patch to webrev? As usual, there's an email attachemnt for your benefit. Thanks.

Le voila: http://cr.openjdk.java.net/~bpb/4511638/webrev.04/

Best regards,


More information about the core-libs-dev mailing list