Hi Phil, As Charles already mentioned this is a known issue. The solution we have in mind is to provide a flag that will allow you to use explicit locking, without scanning of object graphs etc.
This is the next thing I'm going to implement. I hope to get it done within a the next 2 weeks. thanks for your patience, Thomas > -----Urspr�ngliche Nachricht----- > Von: Phil Warrick [mailto:[EMAIL PROTECTED]] > Gesendet: Dienstag, 10. September 2002 18:25 > An: OJB Users List > Betreff: ODMG and locking > > > Hello, > > I've run into the problem of ODMG locking overhead and would > like some > ideas about best OJB practices to deal with it (~4 minutes to > retrieve > one object means something has to be done). I've seen bits > and pieces > of discussion related to this issue, so forgive me if I'm bringing up > old stuff. > > If I understand correctly, enforcing persistence by > reachability means > that once one object is retrieved, this object and all others in the > associated graph are locked. > > Does the locking extend to the contents of proxy references and proxy > collections? > > What are the best ways to limit the extent of the locking in a large > graph? For example, say some classes have instances that are > created/updated often and these classes refers to other > classes that are > much more static. If the graph of the static objects is > large, locking > is a big problem. How can one keep the link between these > classes and > not suffer the lock overhead? > > Your responses would be most appreciated. > > Phil > > > -- > To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
