[hibernate-dev] Shortcomings in org.hibernate.id.enhanced.TableGenerator class

2008-04-27 Thread Tom Muldoon
I apologize if this is a 2nd posting. From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tom Muldoon Sent: Thursday, April 24, 2008 8:04 AM To: 'hibernate-dev@lists.jboss.org' Subject: [hibernate-dev] Shortcomings in org.hibernate.id.enhanced.TableGene

Re: [hibernate-dev] Hibernate Search: Master/Slave DirectoryProviders improvements proposal

2008-04-27 Thread Sanne Grinovero
Hi Hardy, A -- nice I'm getting the chance to teach something to you gurus.. yes this is really required, locking is fine, but there are two problems there; a) visibility: JVM makes no guarantee another thread will see the changes made to a variable by another thread; it could see it "later on"

Re: [hibernate-dev] Hibernate Search: Master/Slave DirectoryProviders improvements proposal

2008-04-27 Thread Hardy Ferentschik
Hi Sanne, A) ... but all comunication beetween the client thread calling "getDirectory()" and the CopyDirectory which defines the current directory is unprotected; also volatile isn't enough to fix it, IMHO an AtomicReference would be more appropriate. Is this really required? Is synchron