[ https://issues.jboss.org/browse/TEIID-2584?page=com.atlassian.jira.plugin... ] Steven Hawkins edited comment on TEIID-2584 at 9/5/13 8:19 AM: --------------------------------------------------------------- I think for now I'm going to move the start/stop stuff off of the metadata api and just make it a concern of the materialization manager. I do see that the tokenize method does handling escaping, so that's not a concern. I'm also going to remove the onErrorAction from the status table and just rely on the one that's on the metadata and when it is set to null/ignore will allow the optimizer to omit the criteria check altogether. Later if we refine the status logic to be something that we hold more or less in memory, then we can retarget the logic of the mvstatus function and not incur the database read. I'll also tweak the mvstatus definition to use the TeiidFunction annotation, which makes it easier to define that the old style SystemSource logic and mark it as non-deterministic so that it will cause the subquery to be reevaluated. Another concern is the approach of using Teiid procedures with XA transactions against sources that require a commit when executing DDL - such as Oracle. I believe that they will simply not allow DDL under an XA transaction. At least for Oracle the caveat is that the scripts may just need to be a little more complicated and use an anonymous subtransaction. was (Author: shawkins): I think for now I'm going to move the start/stop stuff off of the metadata api and just make it a concern of the materialization manager. I do see that the tokenize method does handling escaping, so that's not a concern. I'm also going to remove the onErrorAction from the status table and just rely on the one that's on the metadata and when it is set to null/ignore will allow the optimizer to omit the criteria check altogether. Later if we refine the status logic to be something that we hold more or less in memory, then we can retarget the logic of the mvstatus function and not incur the database read. I'll also tweak the mvstatus definition to use the TeiidFunction annotation, which makes it easier to define that the old style SystemSource logic and mark it as non-deterministic so that it will cause the subquery to be reevaluated. Another concern is the approach of using Teiid procedures with XA transactions against sources that require a commit when executing DDL - such as Oracle. I believe that they will simply not allow DDL under an XA transaction. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira