Add runtimeArgs support to native-maven-plugin

[linghengqian]

• Fixes native-maven-plugin lacks runtimeArgs configuration item similar to native-gradle-plugin #685 .
• Fixes native-maven-plugin cannot publish to local repository on Windows with Simplified Chinese #735 .
• Add runtimeArgs support to native-maven-plugin.
• Support publishing native-maven-plugin to local repository on Windows with Simplified Chinese.
• I have tested this locally with ./gradlew publishToMavenLocal --no-parallel against https://github.com/linghengqian/jetcd-multi-platform-native-test and org.graalvm.buildtools:native-maven-plugin:0.11.0-SNAPSHOT. This even fixes the bug reported in [Native Image] Netty DNS Resolver is not available in GraalVM Native Image compiled on Windows oracle/graal#11280 where the unit tests never stopped.
• 

[alvarosanchez]

With regards to testing this with Spock, you can use @Requires({ jvm.java23Compatible }) but apart from this, is there any other runtime arg that you can test with?

[linghengqian]

is there any other runtime arg that you can test with?

• I would say I have no idea. The GraalVM doc only mentions this one available runtime arg.

With regards to testing this with Spock, you can use @requires({ jvm.java23Compatible }) but apart from this,

• I submitted a new commit to supplement the unit tests, but there is still an awkward problem. It seems that the gradle wrapper 7.4 used by the native-maven-plugin module does not support the JDK 23 runtime. I say this is awkward because I'm testing on Windows and can't set a session-level environment variable GRAALVM_HOME like Ubuntu can.

[alvarosanchez]

@dnestoro please help @linghengqian to write a test for this. The one added here is still skipped.

Other than that, this LGTM

@linghengqian thank you for your patience and collaboration!

[alvarosanchez]

@dnestoro will this go into 0.11.0? If not, please adjust it

[dnestoro]

The one added here is still skipped

I believe that is because the test is annotated with @requires({ jvm.java23Compatible }) (and we are running our dev builds on JDK25).

@linghengqian sorry for the late response. Can you try something like this. You can calculate currently used version like this:

def getCurrentJDKVersion() {
        return NativeImageUtils.getMajorJDKVersion(GraalVMSupport.getGraalVMHomeVersionString())
}

and then your requires annotation should be something like

@Requires({ getCurrentJDKVersion() >= 23 })

[linghengqian]

• After I set GRAALVM_HOME locally to 24.0.1-graalce of vfox, the results of ./gradlew :native-maven-plugin:functionalTest are normal.

• 

• I will say that I am curious about the failed CI involving JDK 26 at https://github.com/graalvm/native-build-tools/actions/runs/15781723681/job/44488746828?pr=734 , mainly I am not sure if SDKMAN! or foojayio/discoapi have JDK26 . vfox install java@26.ea.1-graal doesn't work, which seems to mean https://github.com/graalvm/oracle-graalvm-ea-builds/releases/tag/jdk-26.0.0-ea.01 hasn't been uploaded to foojayio/discoapi yet?

• update in 2025.06.21: I see https://api.foojay.io/swagger-ui#/default/getDistributionsV3 is already documented for JDK 26. I'll investigate later on the vfox side.

• 

[linghengqian]

sorry for the late response. Can you try something like this. You can calculate currently used version like this:

• It seems that starting from GraalVM CE Dev For JDK 25, in the printed Log, org.graalvm.nativeimage.MissingReflectionRegistrationError changed to com.oracle.svm.core.jdk.resources.MissingReflectionRegistrationError, I am not sure which commit of https://github.com/oracle/graal/blob/master/substratevm/src/com.oracle.svm.core/src/com/oracle/svm/core/MissingRegistrationUtils.java caused this change. I changed the assertion slightly.
• All done.

[dnestoro]

@linghengqian thanks for your contribution. From testing side it looks okay now. Since you force pushed commits, original approval from @alvarosanchez is dismissed.

@alvarosanchez can you please re-approve the PR if it is still okay (since I don't know the previous state of PR and there is only one commit in history)