A first version of a Maven profile to run the TCK against Glassfish with Krazo installed as a Glassfish module.
However, the TCK currently fails horribly. It looks like HK2 isn't able to do any CDI injections into our JAX-RS providers (like ViewableWriter). Also, I'm not seeing any logs which indicate that Krazo is bootstrapped correctly at all.
For a quick test, I added a beans.xml to krazo-core with discovery mode all. In this case it seems to work better, but Glassfish reports that our ServletContainerInitializer implementation cannot be loaded. So it looks like Glassfish finds our ServiceLoader file, but fails to load our implementation.
All of this is really weird. And IIRC it worked much better with Glassfish 5 back then.
Here a few tips on how to do similar experiments. Maybe somebody finds some time to have a look as well.
# Build Krazo
cd ${KRAZO_REPO}
mvn -Pstaging -DskipTests clean install
# Copy Krazo JARs + the spec JAR to the Glassfish module directory:
cp core/target/krazo-core-*.jar ${GLASSFISH_HOME}/glassfish/modules/
cp jersey/target/krazo-jersey-*.jar ${GLASSFISH_HOME}/glassfish/modules/
mvn dependency:copy -Dartifact=jakarta.mvc:jakarta.mvc-api:2.0.0 -DoutputDirectory=${GLASSFISH_HOME}/glassfish/modules/
# Start Glassfish
cd ${GLASSFISH_HOME}
./glassfish/bin/startserv
# Run the TCK
cd ${KRAZO_REPO}/tck
mvn -Pstaging -Dtck-env=glassfish-module verify
A first version of a Maven profile to run the TCK against Glassfish with Krazo installed as a Glassfish module.
However, the TCK currently fails horribly. It looks like HK2 isn't able to do any CDI injections into our JAX-RS providers (like
ViewableWriter). Also, I'm not seeing any logs which indicate that Krazo is bootstrapped correctly at all.For a quick test, I added a
beans.xmltokrazo-corewith discovery modeall. In this case it seems to work better, but Glassfish reports that ourServletContainerInitializerimplementation cannot be loaded. So it looks like Glassfish finds our ServiceLoader file, but fails to load our implementation.All of this is really weird. And IIRC it worked much better with Glassfish 5 back then.
Here a few tips on how to do similar experiments. Maybe somebody finds some time to have a look as well.
However, IMO this PR can be already merged.