Re: [eclipselink-users] Specifying CASCADE-REMOVE in project.xml

["Miguel Morales" <rmmora159@xxxxxxxxxxx>]

Tom,

just started developing with dynamic persistence last week, doing very well, 
thanks.  Since you know this stuff so well, can you confirm that if I want 
to solve the cascade issue by using orm or eclipselink-orm I would have to 
specify the entire descriptor info and not just the cascade option ?  Any 
chance it will read the descriptor info from project.xml and apply a single 
attribute (CASCADE) it finds in either of the two previous xml files 
mentioned?  That would be the best solution.

Miguel

--------------------------------------------------
From: "Tom Ware" <tom.ware@xxxxxxxxxx>
Sent: Monday, November 08, 2010 9:29 PM
To: "EclipseLink User Discussions" <eclipselink-users@xxxxxxxxxxx>
Subject: Re: [eclipselink-users] Specifying CASCADE-REMOVE in project.xml

> Hi Miguel,
>
>   You are correct that if you have a large amount of metadata already 
> mapped, the path of least resistance will likely be using the preLogin 
> event.
>
>   The other option, as you have remarked is a conversion to 
> eclipselink-orm.xml.    It covers the majority of the features available 
> through project.xml and also covers the new features introduced in JPA. 
> It will be the main xml-based target for new configuration options we add 
> in the future.  In JPA, you get alot for free because of the defaulting, 
> so depending on how your model is designed you get alot for free.
>
>   BTW: In case you are not aware, we have a feature that allows dynamic 
> generation of Entities.  Here's an example: 
> http://wiki.eclipse.org/EclipseLink/Examples/JPA/Dynamic
>
> -Tom
>
> Miguel Morales wrote:
> > Tom,
> >
> > the Eclipselink Workbench seemed like a more practical interface than 
> > Dali in Eclipse, although later I discovered that it has missing 
> > features.  But I simply cannot afford embedded annotations because my 
> > application will be generating dynamic entities and database features, 
> > and I want to keep all configuration outside of the code itself.  I was 
> > just considering the ORM.XML option when I saw your email, as it is 
> > possible to override configurations using that file and the 
> > eclipselink-orm.xml (something like that).  But it would force me to 
> > specify the entire configuration for that entity just to be able to add 
> > the cascade feature I need.  As much as possible I would like a single 
> > point for generating all necessary JPA stuff, the eclipse workbench, but 
> > barring that I want the least involved option, which at this point I 
> > believe would be your suggestion to use the preLogin. I will look into 
> > it, thank you.
> >
> > Miguel
> > --------------------------------------------------
> > From: "Tom Ware" <tom.ware@xxxxxxxxxx>
> > Sent: Monday, November 08, 2010 8:42 PM
> > To: "EclipseLink User Discussions" <eclipselink-users@xxxxxxxxxxx>
> > Subject: Re: [eclipselink-users] Specifying CASCADE-REMOVE in project.xml
> >
> > > Is your objection to annotations, or to JPA metadata in general?
> > >
> > > If you are are looking for an xml-based way of specifying your metadata, 
> > > JPA allows you to fully specify the metadata in orm.xml files.  You can 
> > > use the Eclipse Dali project for tooling.  orm.xml provides equivalent 
> > > functionality to what is available in annotations.
> > >
> > > If you would like to keep using project.xml, your best bet is to modify 
> > > your mappings through API in an event.  I suggest using the preLogin 
> > > event to modify your descriptors.  You can call setCascadeRemove(true) 
> > > on each ForeignReferenceMapping you would like cascade remove to operate 
> > > on.
> > >
> > > -Tom
> > >
> > > rmmora159 wrote:
> > > > I need a way to specify a CASCADE-REMOVE setting on a one-to-many
> > > > relationship in the project.xml of my eclipselink app.  Currently a 
> > > > remove()
> > > > on the parent entity throws an SQL exception because of the foreign key
> > > > constraint.  I know this is easily solved using annotations,  but I DO 
> > > > NOT
> > > > use annotations nor will I use them because they are a waste of time. 
> > > > It is
> > > > much more efficient to use a single project.xml that describes 
> > > > everything. Furthermore, my application generates its own database 
> > > > tables and entities,
> > > > so I need to keep everything as centralized and hands-free as possible. 
> > > > I
> > > > use the eclipse Workbench (I know, I shouldn't) to generate all the
> > > > necessary components, but that GUI does not seem to have an option to 
> > > > set
> > > > cascading type for relationships.  I know I'll probably have to 
> > > > manually add
> > > > the CASCADE setting to the project.xml, but I cannot figure out the xsd
> > > > specifications to correctly insert it into the xml. I could also take 
> > > > care
> > > > of the problem in java before calling the remove() method of the
> > > > EntityManager, if someone knows how to do so.
> > > _______________________________________________
> > > eclipselink-users mailing list
> > > eclipselink-users@xxxxxxxxxxx
> > > https://dev.eclipse.org/mailman/listinfo/eclipselink-users
> > _______________________________________________
> > eclipselink-users mailing list
> > eclipselink-users@xxxxxxxxxxx
> > https://dev.eclipse.org/mailman/listinfo/eclipselink-users
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users