Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.Sign up
Query clone support with pre-existing entityManager/connection #672
I've used so far the jpaSqlQuery.clone(entityManager) method and thought to give it a try on the sql side as well. First of all it was missing from the mysql specific implementation altogether, but luckily it was present on the more generic SQLQuery.
On the JPA side it was easy enough (although a bit irritating) to carry around the entity manager as well but to get a hold to the same connection that 's inside the original sqlQuery proved to be problematic.
Would it be possible to introduce a new clone() method on the interface level and reintroduce the pre-existing clone(entityManager/connection) as abstract on both of the AbstractSQLQuery implementations. This would also ensure that all of the implementations implement also the cloning.
Here is a summary of what you suggest:
In addition to this implementation of the missing clone methods in the subclasses.
The interfaces for SQL queries, e.g. SQLCommonQuery, are more meant to enforce consistent naming for queries and subqueries.
Could you confirm whether I got it right?
That's the part one. I think that AbstractSQLSubQuery could also benefit from cloning. As a side note: SQLServerSubQuery doesn't seem to be included to SQLServerQueryFactory).
The second part of the request would be to clone without knowing anything that is laying below the bonnet (my original use case). I know this suggestion contradicts a bit what you stated but due to the lack of common ancestor class this is the best that I came up with.
com.mysema.query.sql.SQLCommonQuery<Q extends SQLCommonQuery<Q>> public Q clone();
Sorry about the long delay. Here a pull request #713
It wasn't as straight forward as I'd first though and I guess I over-accomplished a bit by pulling out common ancestors for both detached and projectable sql queries so that I could get rid of clear duplicates (sql/jpa.AbstractSQLQuery -> ProjectableSQLQuery and sql.AbstractSQLSubQuery -> DetachableSQLQuery). I also refactored out constants from previous serializer instantiation that were preserved among internal state.
I guess you were referring with the semantics to this change that you posted to me in another issue 69ea056
I chose the diplomatic route and used the same approach although I think it would be wiser to check if the templating is the default one and only then apply the JPAProvider.getTemplates(). With the current approach it would be impossible to use custom queryhandler coherently since it would be lost in the cloning process.