If the wrapper is driver specific we could implement autonumber fields very easily, with one specific implementation per vendor. I don't know when a MySql JDBC 3.0 driver will be available. JDBC has a common way to obtain the value of a autonumer field, after insertion. Since JDBC 3.0 drivers (specifically for MySql, maybe for other databses too) are not ready yet, we could use the wrapper to access vendor specific funcionality.
This could help a lot of people having trouble with object ID generation... And also people working with existing databases, with autonumber primary keys. What do you think? Ricardo Arguello David Jencks writes: > I believe the problem is in the jca jdbc wrappers, not the > ConnectionManager or ConnectionEventListener implementations. The current > code in 3.0 appears to not call ConnectionErrorOccurred no matter what > SQLException is thrown. > > My opinion is that the wrapper should be driver specific and know whether a > particular sql exception means the connection is dead, or perhaps simply an > invalid sql statement was executed. What do you think? > > david jencks > > On 2002.03.18 16:33:15 -0500 Ruwei Hu wrote: >> In my concurrent stress tests, I noticed dead connections are not cleaned >> properly. Brief searching in the JBoss mail list archieve shows similar >> experience from other users. I am using JBoss 2.4.1 with the >> ConnectionFactoryLoader and InvalidateOnError set to true. >> >> Is the issue get fixed in later version of JBoss pool implementation? >> >> Thanks for any information. >> >> Randy >> _______________________________________________ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
