MappingException

[smogstate]

Hello,

it seems we getting same kind of symptoms described here: #252 , #248. We got multiple MapperFactories and Mappers so they are not singletones.

Sometimes really wired exceptions happens like:
ma.glasnost.orika.MappingException: exception while creating object factory for org.joda.time.chrono.ISOChronology at ma.glasnost.orika.impl.generator.ObjectFactoryGenerator.build(ObjectFactoryGenerator.java:104) at ma.glasnost.orika.impl.DefaultMapperFactory.lookupObjectFactory(DefaultMapperFactory.java:966) at ma.glasnost.orika.impl.MapperFacadeImpl.resolveMappingStrategy(MapperFacadeImpl.java:182) at ma.glasnost.orika.impl.DefaultBoundMapperFacade$BoundStrategyCache.getStrategy(DefaultBoundMapperFacade.java:259) at ma.glasnost.orika.impl.DefaultBoundMapperFacade.map(DefaultBoundMapperFacade.java:137) at ma.glasnost.orika.generated.DateTime_DateTime_ObjectFactory43244272000899444324688331862560$9984.create(DateTime_DateTime_ObjectFactory43244272000899444324688331862560$9984.java)

But we defined DateTime as pass through why the mapper is generated?

MapperFactory mapperFactory = new DefaultMapperFactory.Builder()
        .build();
    mapperFactory
        .getConverterFactory()
        .registerConverter(
            new PassThroughConverter(
                LocalDateTime.class, DateTime.class, DateTimeZone.class, ISOChronology.class));
    return mapperFactory;

I checked the implementation of JavassistCompilerStrategy there are couple questions:

1. why superClasses map is static
2. why in registerClassLoader function synchronization is done on parameter

Those exception ofc happens only in production. i cant reproduce it locally even with high concurrent simulation. Any ideas what can go wrong?

we are currently using version 1.5.0.

[smogstate]

Problem solved by instantiating all mappers in single threaded environment. I am strongly recommend to avoid dynamic mapper generation in high conccurent environment.