Hi, Mwanji Your observations about transactions are correct. A while ago I removed the transaction check for read operations (as a performance optimization). Reading may however throw an exception depending on what state the object being read is in.
Currently all node, relationship and property operation should be performed in a transaction but we are having discussions about removing the transaction constraint for read operations. -Johan On Sun, Dec 28, 2008 at 6:16 PM, Mwanji Ezana <mwa...@gmail.com> wrote: > Hi, > While writing some unit tests, I realised that I was performing some > read-only assertions without being in a transaction. Exploring the issue a > little more, it seems that nodes accessed within a transaction can then be > accessed from outside a transaction. Is this the case? Or is my design > "leaking" something? > > I've wrapped try/finally programmatic transaction management in an abstract > class which is anonymously subclassed to perform Neo operations. My actual > domain objects don't deal with transactions at all, but run in transactions > started at the service/controller level. > > For example, I have reference -> Domains -> Domain -> Projects ->Project. If > I access Project1 in a transaction, I can later access it without a > transaction, but I can't access Project2. > > I'm not sure if I'm doing something wrong or if this behaviour is normal. (I > hope my explanation is clear enough...) > > Mwanji _______________________________________________ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
