Hi again! Wednesday 08 October 2008 15:57:20 Christophe Lombart написав: > Personally, I like the rule 7 made by David. > http://wiki.apache.org/jackrabbit/DavidsModel#head-79ca455cb5a0f371923912cc >1418831f0e68f674 > > In all cases, OCM can manage UUID but please understand that many content > applications can live without that. > Yes I like this rule too and that's why i do not have IDs in my applications, I use UUIDs and references like this rule #7 recommeds. :) And, by the way, examples in your tests that work with references. No way I do not insist on using UUID in OCM code some other way it already implemented. I quite happy with (uuid=yes) and ocm.getObjectByUUID().
Only one thing I wish is Map<AType,AType> implementation. But UUID is nothing about it. :) May be I wish somewhat stricter type checking for ocm.getObject() but I do not know how implement it. And may be I do not wish to register nodes in ocm registry, this could be done automatically by @Node annotation. What I tried to explain is importance of UUID for reference integrity of key. And if someone is going to implement JPA annotations for JCR, UUID will be only one right solution for primary key. -- SY, Alex Lukin RIPE NIC HDL: LEXA1-RIPE
