Potential regression from 0.2.5 to 0.2.6 with label processing

[ingmarfjolla]

Hello, I'm not sure if perhaps something changed that requires extra configuration, but it seems that upgrading to 0.2.6 changes the way that labels/flags are processed. Working on this issue for Quarkus Superheroes, I'm seeing different output in my values yaml and what gets placed in the templates as well. I attached some pictures to make it easier to show what I'm talking about. The command that's being run when building the app is:

/mvnw -f $project/pom.xml versions:set clean package \
    -DskipTests \
    -DnewVersion=$container_tag \
    -Dmaven.compiler.release=$javaVersion \
    -Dquarkus.container-image.tag=$container_tag \
    -Dquarkus.kubernetes.deployment-target=$deployment_type \
    -Dquarkus.kubernetes.version=$container_tag \
    -Dquarkus.kubernetes.ingress.expose=$expose \
    -Dquarkus.kubernetes.resources.limits.memory=$mem_limit \
    -Dquarkus.kubernetes.resources.requests.memory=$mem_request \
    -Dquarkus.kubernetes.annotations.\"app.quarkus.io/vcs-url\"=$GITHUB_SERVER_URL/$GITHUB_REPOSITORY \
    -Dquarkus.kubernetes.annotations.\"app.quarkus.io/vcs-ref\"=$github_ref_name \
    -Dquarkus.openshift.version=$container_tag \
    -Dquarkus.openshift.route.expose=$expose \
    -Dquarkus.openshift.resources.limits.memory=$mem_limit \
    -Dquarkus.openshift.resources.requests.memory=$mem_request \
    -Dquarkus.openshift.annotations.\"app.openshift.io/vcs-url\"=$GITHUB_SERVER_URL/$GITHUB_REPOSITORY \
    -Dquarkus.openshift.annotations.\"app.openshift.io/vcs-ref\"=$github_ref_name \
    -Dquarkus.knative.version=$container_tag \
    -Dquarkus.knative.labels.\"app.openshift.io/runtime\"=quarkus \
    -Dquarkus.knative.resources.limits.memory=$mem_limit \
    -Dquarkus.knative.resources.requests.memory=$mem_request \
    -Dquarkus.knative.annotations.\"app.openshift.io/vcs-url\"=$GITHUB_SERVER_URL/$GITHUB_REPOSITORY \
    -Dquarkus.knative.annotations.\"app.openshift.io/vcs-ref\"=$github_ref_name

the ones with the extra values that are empty are with 0.2.6.

[Sgitario]

This is because we're exposing all the system properties you use as params, so you can change it when installing the chart.
Though, are the empty properties causing failures when installing the chart? Otherwise, I don't see it as a potential regression. You can always disable this feature by adding the property: quarkus.helm.map-system-properties=false.

Moreover, I will make a change to ignore the quarkus.kubernetes.xxx properties.