Hi all, Le 25/02/2015 20:24, Thomas Neidhart a écrit : > On 02/25/2015 04:37 PM, Phil Steitz wrote: >> Seems a little silly to be spending time debugging openjdk bugs. >> Can we just set Jenkins up to use an Oracle released JDK? > > Yeah sure, it might be a waste of time, it is still annoying to see > these test failures and they might also happen in the wild I guess.
First sorry for my current lack of responsiveness, I intend to be back next week. I'm not sure if it is related, but several coworkers at office experienced failing tests too dependeing on the way they compiled some libraries depending heavily on [math]. All these failures seems related to small numerical discrepencies and it seemed to depend on the compilation toolchain (Eclipse or maven and various JDKs). I think almost any cases that were reported to me were run on Linux boxes (Debian or RedHat), but perhaps one or two were on windows (don't know the OS version). So I'm not sure the problems only affect our continuous integration tests, they may well be experienced by users. best regards, Luc > > The information about the installed jdks on the build slaves is actually > very sparse, but from the system information I could derive that the > official Oracle JDK 1.7.0_55 is installed. > > Thomas > >> >> On 2/25/15 2:01 AM, t...@apache.org wrote: >>> Repository: commons-math >>> Updated Branches: >>> refs/heads/master 6b1b59288 -> ccd0fd51a >>> >>> >>> Temp changes for analyzing test failures. >>> >>> >>> Project: http://git-wip-us.apache.org/repos/asf/commons-math/repo >>> Commit: http://git-wip-us.apache.org/repos/asf/commons-math/commit/ccd0fd51 >>> Tree: http://git-wip-us.apache.org/repos/asf/commons-math/tree/ccd0fd51 >>> Diff: http://git-wip-us.apache.org/repos/asf/commons-math/diff/ccd0fd51 >>> >>> Branch: refs/heads/master >>> Commit: ccd0fd51a5712ef1d85110eb61130c906fe4db13 >>> Parents: 6b1b592 >>> Author: tn <thomas.neidh...@gmail.com> >>> Authored: Wed Feb 25 10:01:02 2015 +0100 >>> Committer: tn <thomas.neidh...@gmail.com> >>> Committed: Wed Feb 25 10:01:02 2015 +0100 >>> >>> ---------------------------------------------------------------------- >>> src/main/java/org/apache/commons/math4/util/FastMath.java | 10 ++++++++-- >>> 1 file changed, 8 insertions(+), 2 deletions(-) >>> ---------------------------------------------------------------------- >>> >>> >>> http://git-wip-us.apache.org/repos/asf/commons-math/blob/ccd0fd51/src/main/java/org/apache/commons/math4/util/FastMath.java >>> ---------------------------------------------------------------------- >>> diff --git a/src/main/java/org/apache/commons/math4/util/FastMath.java >>> b/src/main/java/org/apache/commons/math4/util/FastMath.java >>> index 015dbcb..052c7eb 100644 >>> --- a/src/main/java/org/apache/commons/math4/util/FastMath.java >>> +++ b/src/main/java/org/apache/commons/math4/util/FastMath.java >>> @@ -1468,6 +1468,10 @@ public class FastMath { >>> return x; >>> } >>> >>> + // special case #3 described in javadoc of Math.pow(double, double) >>> + if (y != y) { // Y is NaN >>> + return y; >>> + } >>> >>> if (x == 0) { >>> long bits = Double.doubleToRawLongBits(x); >>> @@ -1510,7 +1514,8 @@ public class FastMath { >>> return Double.NaN; >>> } >>> >>> - if (x * x > 1.0) { >>> + // TEMP: was x * x > 1.0 >>> + if (abs(x) > 1.0) { >>> return Double.POSITIVE_INFINITY; >>> } else { >>> return 0.0; >>> @@ -1547,7 +1552,8 @@ public class FastMath { >>> return Double.NaN; >>> } >>> >>> - if (x * x < 1.0) { >>> + // TEMP: was x * x < 1.0 >>> + if (abs(x) < 1.0) { >>> return Double.POSITIVE_INFINITY; >>> } else { >>> return 0.0; >>> >>> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
