On Sat, 18 Apr 2026 at 14:15, Gilles Sadowski <[email protected]> wrote:
> Hello. > > Le jeu. 16 avr. 2026 à 14:13, Alex Herbert <[email protected]> a > écrit : > > > > We have fixed quite a few bugs and added some significant enhancements > > since Apache Commons RNG 1.6 was released, so I would like to release > > Apache Commons RNG 1.7. > > > > Apache Commons RNG 1.7 RC1 is available for review here: > > https://dist.apache.org/repos/dist/dev/commons/rng/1.7-RC1 (svn > > revision 83845) > > > > The Git tag commit for this RC is commons-rng-1.7-RC1, which you can > browse > > here: > > > > > https://gitbox.apache.org/repos/asf?p=commons-rng.git;a=commit;h=commons-rng-1.7-RC1 > > You may checkout this tag using: > > git clone https://gitbox.apache.org/repos/asf/commons-rng.git > --branch > > commons-rng-1.7-RC1 commons-rng-1.7-RC1 > > (signature can be checked from git using 'git tag -v > commons-rng-1.7-RC1') > > After a fresh "git clone ...": > $ mvn > [... skipped ...] > [INFO] > ------------------------------------------------------------------------ > [INFO] Reactor Summary for Apache Commons RNG 1.7: > [INFO] > [INFO] Apache Commons RNG ................................. SUCCESS [ > 14.753 s] > [INFO] Apache Commons RNG Client API ...................... FAILURE [ > 26.094 s] > [INFO] Apache Commons RNG Core ............................ SKIPPED > [INFO] Apache Commons RNG Simple .......................... SKIPPED > [INFO] Apache Commons RNG Sampling ........................ SKIPPED > [INFO] Apache Commons RNG (Bill of Materials) ............. SKIPPED > [INFO] Apache Commons RNG Documentation ................... SKIPPED > [INFO] > ------------------------------------------------------------------------ > [INFO] BUILD FAILURE > [INFO] > ------------------------------------------------------------------------ > [INFO] Total time: 41.312 s > [INFO] Finished at: 2026-04-18T14:29:29+02:00 > [INFO] > ------------------------------------------------------------------------ > [ERROR] Failed to execute goal > org.revapi:revapi-maven-plugin:0.15.1:check (default) on project > commons-rng-client-api: Execution default of goal > org.revapi:revapi-maven-plugin:0.15.1:check failed: A required class > was missing while executing > org.revapi:revapi-maven-plugin:0.15.1:check: > org/apache/logging/log4j/spi/LoggerAdapter > [ERROR] ----------------------------------------------------- > [ERROR] realm = plugin>org.revapi:revapi-maven-plugin:0.15.1 > [ERROR] strategy = > org.codehaus.plexus.classworlds.strategy.SelfFirstStrategy > [ERROR] urls[0] = > > file:/home/gilles/.m2/repository/org/revapi/revapi-maven-plugin/0.15.1/revapi-maven-plugin-0.15.1.jar > [... skipped ...] > [ERROR] urls[60] = > > file:/home/gilles/.m2/repository/com/ximpleware/vtd-xml/2.11/vtd-xml-2.11.jar > [ERROR] Number of foreign imports: 1 > [ERROR] import: Entry[import from realm ClassRealm[maven.api, parent: > null]] > [ERROR] > [ERROR] -----------------------------------------------------: > org.apache.logging.log4j.spi.LoggerAdapter > [ERROR] -> [Help 1] > [...] > Strange errors about log4j. We do not use this. It seems to be missing from your revapi plugin dependency stack. I wonder if the dependencies are different on your machine to mine. I built the release on a Mac workstation. I've just verified the release on my laptop (also a Mac) and the build is ok: 504 git clone https://gitbox.apache.org/repos/asf/commons-rng.git --branch commons-rng-1.7-RC1 commons-rng-1.7-RC1 505 cd commons-rng-1.7-RC1/ 507 mvn # Build is OK 508 mvn -v Apache Maven 3.9.4 (dfbb324ad4a7c8fb0bf182e6d91b0ae20e3d2dd9) Maven home: /Users/ah403/mvn/mvn Java version: 17.0.17, vendor: Eclipse Adoptium, runtime: /Library/Java/JavaVirtualMachines/temurin-17.jdk/Contents/Home Default locale: en_GB, platform encoding: UTF-8 OS name: "mac os x", version: "26.3.1", arch: "aarch64", family: "mac" Note that 'mvn clean install' also works for me on the same Macbook. > > Running, in the same directory, > $ mvn clean install > [... skipped ...] > [INFO] > ------------------------------------------------------------------------ > [INFO] Reactor Summary for Apache Commons RNG 1.7: > [INFO] > [INFO] Apache Commons RNG ................................. SUCCESS [ > 11.970 s] > [INFO] Apache Commons RNG Client API ...................... SUCCESS [ > 18.703 s] > [INFO] Apache Commons RNG Core ............................ FAILURE [ > 55.527 s] > [INFO] Apache Commons RNG Simple .......................... SKIPPED > [INFO] Apache Commons RNG Sampling ........................ SKIPPED > [INFO] Apache Commons RNG (Bill of Materials) ............. SKIPPED > [INFO] Apache Commons RNG Documentation ................... SKIPPED > [INFO] > ------------------------------------------------------------------------ > [INFO] BUILD FAILURE > [INFO] > ------------------------------------------------------------------------ > [INFO] Total time: 01:26 min > [INFO] Finished at: 2026-04-18T14:39:25+02:00 > [INFO] > ------------------------------------------------------------------------ > [ERROR] Failed to execute goal > org.revapi:revapi-maven-plugin:0.15.1:check (default) on project > commons-rng-core: The following API problems caused the build to fail: > [ERROR] java.class.externalClassExposedInAPI: interface > org.apache.commons.rng.JumpableUniformRandomProvider: A class from > supplementary archives is used in a public capacity in the API. > [org.apache.commons:commons-rng-client-api:jar:1.7] > [ERROR] java.class.externalClassExposedInAPI: interface > org.apache.commons.rng.RandomProviderState: A class from supplementary > archives is used in a public capacity in the API. > [org.apache.commons:commons-rng-client-api:jar:1.7] > [ERROR] java.class.externalClassExposedInAPI: interface > org.apache.commons.rng.UniformRandomProvider: A class from > supplementary archives is used in a public capacity in the API. > [org.apache.commons:commons-rng-client-api:jar:1.7] > [...] > > $ mvn --version > Apache Maven 3.9.9 > Maven home: /usr/share/maven > Java version: 21.0.10, vendor: Debian, runtime: > /usr/lib/jvm/java-21-openjdk-amd64 > Default locale: en_US, platform encoding: UTF-8 > OS name: "linux", version: "6.12.74+deb13+1-amd64", arch: "amd64", > family: "unix" > This is a strange error message. The core component has always used interfaces from the client-api component. I do not understand why these should fail now if they did not in the past. Note that we have to add new interfaces from the client-api to revapi's ignore configuration. This was done for the new interfaces: SplittableUniformRandomProvider (1.5); ArbitrarilyJumpableUniformRandomProvider (1.7 - new this release). See: src/conf/revapi/api-changes.json Here revapi is complaining about interfaces that have been in the code from at least v1.3: JumpableUniformRandomProvider. I tried on a clean linux platform (fresh maven install of 3.9.15) using: Apache Maven 3.9.15 (98b2cdbfdb5f1ac8781f537ea9acccaed7922349) Maven home: /its/home/ah403/ah403/apache-maven-3.9.15 Java version: 11.0.20, vendor: Eclipse Adoptium, runtime: /mnt/shared/easybuild/software/Java/11.0.20 Default locale: en_US, platform encoding: UTF-8 OS name: "linux", version: "5.14.0-503.16.1.el9_5.x86_64", arch: "amd64", family: "unix" # Use a new local maven repo ../apache-maven-3.9.15/bin/mvn -Dmaven.repo.local=../repository This worked OK. FYI the repository size after this was 325M. This is quite a lot of tooling (excluding maven and the JDK) for a build release of 4.5M for the binary jars and ~3M for the source. > Regards, > Gilles > > P.S. The current build ("CI" label) on GH is marked as "Failing": > https://github.com/apache/commons-rng/actions/workflows/maven.yml It failed due to random tests not achieving the p-threshold. So it is not related. This is increasingly common as we add more and more tests and use either a 0.01 or 0.001 critical value. We allow some repeats for flakes: > git grep rerunFailingTestsCount commons-rng-core/pom.xml: <rerunFailingTestsCount>2</rerunFailingTestsCount> commons-rng-sampling/pom.xml: <rerunFailingTestsCount>1</rerunFailingTestsCount> commons-rng-simple/pom.xml: <rerunFailingTestsCount>1</rerunFailingTestsCount> So in the sampling component we use 1 rerun. The failing test had a p-value of 0.001 and should fail 1 in 1,000,000 times. I've looked at the test and believe it is doing the correct Chi-square test on a contingency table. I added some logging of the parameters for the test to help future debugging. After this push the build is now green again. Alex > > > [...] > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
