[hibernate-dev] Dialect#remapSqlTypeDescriptor

[Steve Ebersole]

Actually I am finding a lot of references online that calling Oracle
procs/functions that define PL/SQL BOOLEAN arguments/returns cannot be done
via CallableStatement.  This is going to take some investigation.

On Fri, Sep 30, 2016 at 7:10 PM Steve Ebersole <steve at hibernate.org> wrote:

> Specifically, does anyone know how Oracle driver
> reports ParameterMetaData#getParameterType for CallableStatement arguments
> that are defined as PL/SQL BOOLEAN types?
>
> On Fri, Sep 30, 2016 at 7:02 PM Steve Ebersole <steve at hibernate.org>
> wrote:
>
> We may be able to look at the procedures Metadata from CallableStatement.
> I forget the exact details of how it reports the argument types, especially
> for Oracle specifcally
>
> On Fri, Sep 30, 2016, 12:57 PM Vlad Mihalcea <mihalcea.vlad at gmail.com>
> wrote:
>
> I also consider that we should provide info so that we can properly map
> the Boolean type according to the expected SQL type.
> Especially since we have revamped the support for Stored procedures and
> functions.
>
> Vlad
> On Fri, Sep 30, 2016 at 5:11 PM, Steve Ebersole <steve at hibernate.org>
> wrote:
>
> TLDR: Should we adjust to allow Dialect to know the "context" of where the
> remapping is requested?
>
> Ah Oracle...
>
> So this comes from the fact that Oracle does not support a BOOLEAN
> datatype.  Well kind of.  It does not support a BOOLEAN datatype in its
> "SQL engine".  However, in PL/SQL it does in fact support a BOOLEAN
> datatype.  Which comes into play when we talk about calling functions and
> procedures: the arguments and returns can in fact be a BOOLEAN.
>
> As far as I know, Oracle is the only database with this type of
> inconsistency in its type system.  But the question is whether we want to
> pass along some kind of information regarding the context (SQL v function v
> procedure) to the Dialect#remapSqlTypeDescriptor.
>
> You can look at https://hibernate.atlassian.net/browse/HHH-11141 for an
> illustration of how this impacts applications.  And if you look through the
> comments you can see the kind-of-crazy workaround needed.
>
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>
>