[infinispan-issues] [JBoss JIRA] Commented: (ISPN-701) Redesign TableManipulation in JDBC cache loader
Manik Surtani (JIRA)
jira-events at lists.jboss.org
Tue Apr 26 07:18:18 EDT 2011
[ https://issues.jboss.org/browse/ISPN-701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12597956#comment-12597956 ]
Manik Surtani commented on ISPN-701:
------------------------------------
@Trustin I presume you will re-jig this JIRA to suit the results of our design discussions via email?
> Redesign TableManipulation in JDBC cache loader
> -----------------------------------------------
>
> Key: ISPN-701
> URL: https://issues.jboss.org/browse/ISPN-701
> Project: Infinispan
> Issue Type: Task
> Components: Loaders and Stores
> Reporter: Trustin Lee
> Assignee: Trustin Lee
> Fix For: 5.0.0.CR1, 5.0.0.FINAL
>
>
> There are two on-going issues related with TableManipulation at the moment: ISPN-686 and ISPN-698. They both are related with vendor specific behavior, and the current implementation uses switch-cases to deal with the differences between vendors. Could we instead use inheritance to make the code look cleaner and easier to maintain? Hibernate does so:
> http://viewvc.jboss.org/cgi-bin/viewvc.cgi/hibernate/core/trunk/core/src/main/java/org/hibernate/dialect/
> Also, the properties like custom types, names, prefixes, fetch/batch sizes could be moved to AbstractJdbcCacheStoreConfig (or its subclass because we have mixed JDBC store) instead of exposing TableManipulation directly to a user.
> Since Hibernate already provides very well defined dialect metadata model, we could simply tap into it. However, we should wrap it with a simple wrapper class so that a user can configure the JDBC store without the knowledge of Hibernate.
> This is a backward incompatible change - will be done in 5.0, and TableManipulation and its related methods should be deprecated in 4.2.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
More information about the infinispan-issues
mailing list