The CTE is bound to the currently executing query. It takes the ids that
need otherwise would be in a temporary table and feeds them using the
VALUES() expression.
It's interesting the reason for doing so: "reducing concurrency on system
tables every time Postgresql creates temp tables".
I'd have to do some research regarding that statement.
Does it sound like a good idea to you?
Vlad
On Wed, Mar 23, 2016 at 6:43 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
How does one reference a CTE across multiple deletes via JDBC? A
CTE, by
definition, is available only during the execution of a single query. In
most databases you "work around" that by grouping statements into a block
(GO, BEGIN/END etc). This is what JDBC poorly attempt to mimic with
PreparedStatement batching. So practically I am just not sure how we'd do
this from JDBC. Are you thinking we'd build database-specific blocks?
On Wed, Mar 23, 2016 at 8:05 AM Vlad Mihalcea <mihalcea.vlad(a)gmail.com>
wrote:
> Hi,
>
> I found a very interesting solution for the Bulk id multi-table strategy
> for those cases when the DB management policy forbids creating Temporary
> Tables.
> This situation is documented in this Forum post:
>
>
https://forum.hibernate.org/viewtopic.php?f=1&t=1043080&p=2489096...
>
> After doing some investigation, I found this project on GitHub:
>
>
https://github.com/epiresdasilva/cte-multi-table-bulk-id-stategy
>
> After migrating it for Hibernate 5.x, I tested it and it works very nice,
> generating queries like this:
>
> with HT_announecement ( id ) as (
> values
> (?), (?) )
> delete
> from
> announcement
> where
> (
> id
> ) IN (
> select
> id
> from
> HT_announecement
> )
>
> I think we should take this idea and write and come up with a built-in
> strategy using CTE.
>
> What do you think?
>
> Vlad
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/hibernate-dev
>