[ https://issues.jboss.org/browse/ISPN-7990?page=com.atlassian.jira.plugin.... ] Sanne Grinovero commented on ISPN-7990: --------------------------------------- bq. And in Infinispan, we have the advantage of having unique document ids across all indices (it derives from the cache key). When you mention "unsafe", is it some internal limitation of the Hibernate Search engine, or the QueryKnownClasses (which is backed by a SYNC REPL cache) itself? No I'm not referring to some obscure limitation in Hibernate Search. I was merely concerned about the reliability (especially over time and in exceptional cases) of {{QueryKnownClasses}}, but since Adrian clarified this is finally being taken care of I guess I have no more concerns. Incidentally you remind me of the terrible hack that Infinispan Query does to encode keys. We should see if we can improve on that, especially after we've dropped "auto-discovery" of index mapping. -- This message was sent by Atlassian JIRA (v7.2.3#72005)