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

2021-11-15 Thread Raffaello Giulietti

Hello,

the 2nd version of the article [1] accompanying this issue [2] and issue 
[3] has been kindly and thoroughly reviewed by the following renowned 
world-class researchers:


* Guy Steele Jr [4], Oracle Labs
* Paul Zimmermann [5], INRIA
* Jean-Michel Muller [6], Ecole Normale Supérieure de Lyon

Thanks to the improvements contributed by these fine reviewers, the 
newest 4th version [7] is now in "preview" form, the intent being to 
publish it on a more permanent platform (academic journal or similar).


In addition, they also suggested and provided new hard tests which are 
included in the PR [8].


I hope this will help in progressing both the CSR and the PR.


Greetings
Raffaello



[1] https://drive.google.com/file/d/1luHhyQF9zKlM8yJ1nebU0OgVYhfC6CBN
[2] https://bugs.openjdk.java.net/browse/JDK-8202555
[3] https://bugs.openjdk.java.net/browse/JDK-4511638
[4] https://labs.oracle.com/pls/apex/f?p=labs%3Abio%3A0%3A120
[5] https://members.loria.fr/PZimmermann/
[6] https://perso.ens-lyon.fr/jean-michel.muller/
[7] https://drive.google.com/file/d/1IEeATSVnEE6TkrHlCYNY2GjaraBjOT4f
[8] https://github.com/openjdk/jdk/pull/3402



On 2021-10-26 21:28, Raffaello Giulietti wrote:

Hello,

PR [1] and the accompanying article [2] have been subject to some 
positive reactions in the last couple of weeks. A fresh set of about 20 
thousand additional hard test cases kindly provided by Paul Zimmermann 
of INRIA and other tests proposed by Guy Steele have been added to the 
code.


The corresponding CSR [3] is in Finalized state for review. The proposed 
spec has been carefully written to uniquely define the outcomes and the 
code in the PR has been extensively tested to match the proposed spec.


Behavioral backward compatibility has been a major goal for the CSR. In 
fact, in the vast majority of cases, the CSR and the current 
implementation agree on the outcomes.


As the CSR is a prerequisite for the advancement of the PR, I beg 
everybody entitled (and interested) to review and approve it and/or 
discuss it further.



Greetings
Raffaello



[1] https://github.com/openjdk/jdk/pull/3402
[2] https://drive.google.com/file/d/1luHhyQF9zKlM8yJ1nebU0OgVYhfC6CBN/view
[3] https://bugs.openjdk.java.net/browse/JDK-8202555







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

2021-10-26 Thread Raffaello Giulietti

Hello,

PR [1] and the accompanying article [2] have been subject to some 
positive reactions in the last couple of weeks. A fresh set of about 20 
thousand additional hard test cases kindly provided by Paul Zimmermann 
of INRIA and other tests proposed by Guy Steele have been added to the code.


The corresponding CSR [3] is in Finalized state for review. The proposed 
spec has been carefully written to uniquely define the outcomes and the 
code in the PR has been extensively tested to match the proposed spec.


Behavioral backward compatibility has been a major goal for the CSR. In 
fact, in the vast majority of cases, the CSR and the current 
implementation agree on the outcomes.


As the CSR is a prerequisite for the advancement of the PR, I beg 
everybody entitled (and interested) to review and approve it and/or 
discuss it further.



Greetings
Raffaello



[1] https://github.com/openjdk/jdk/pull/3402
[2] https://drive.google.com/file/d/1luHhyQF9zKlM8yJ1nebU0OgVYhfC6CBN/view
[3] https://bugs.openjdk.java.net/browse/JDK-8202555