Howdy, tried locally, and I think I got it. But, by tweaking your code, I spotted that it is your input that is unstable (different), and hence Resolver comes up with different output as well. Make sure you feed the same input (we talk about the same test case, so project is same) for the same project and then you can expect the same (resolver) output, otherwise Resolver will carry on the "randomization" from your code. Simply put, the project Jack.toml produces different calls to Resolver, but did not dig deep into the reasons (concurrency?)
Thanks T On Fri, Aug 9, 2024 at 11:09 AM Tamás Cservenák <ta...@cservenak.net> wrote: > Howdy, > > How could I reproduce this locally? > > T > > On Sat, Jul 20, 2024 at 1:28 PM Nick Hensel > <nickhense...@icloud.com.invalid> wrote: > >> Hi, >> >> I'm resolving dependencies with the help of Maven Artifact Resolver >> (2.0) and I have an issue where resolved (transitive) dependencies >> differ from run to run. Currently I'm using "new >> ConfigurationVersionSelector(new ConfigurableVersionSelector.Highest())" >> as my VersionSelector. >> >> In most cases the same (transitive) dependencies are resolved, but >> occasionally some dependencies are missing or too much. >> >> In my particular test case >> ( >> https://codeberg.org/goldmensch/jack/src/branch/jdk-installation-checksum-support/test_project/project/Jack.toml) >> >> the transitive dependencies "javax.inject:javax.inject:1" and >> "org.eclipse.sisu:org.eclipse.sisu.inject:0.9.0.M3" are sometimes >> resolved and sometimes not. >> >> The involved code is located here >> >> https://codeberg.org/goldmensch/jack/src/branch/jdk-installation-checksum-support/src/main/java/io/github/goldmensch/dependencies >> . >> >> >> Best regards, >> >> Nick Hensel >> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org >> For additional commands, e-mail: dev-h...@maven.apache.org >> >>
