[hibernate-dev] metamodel work
gbadner at redhat.com
Wed May 23 14:05:19 EDT 2012
----- Original Message -----
> From: "Hardy Ferentschik" <hardy at hibernate.org>
> To: "Hibernate" <hibernate-dev at lists.jboss.org>
> Sent: Wednesday, May 23, 2012 3:34:38 AM
> Subject: Re: [hibernate-dev] metamodel work
> On May 22, 2012, at 11:41 PM, Steve Ebersole wrote:
> > This transition from "Configuration metadata" to "new Metadata" is
> > just
> > that.. a transition. I'd really rather not spend a lot of time
> > developing testing extensions for dealing with the transitory
> > nature of
> > this.
> I also cannot imagine that we can keep both code paths (new and old
> working for ever. At some stage we will have to wire up the new stuff
> for good and I don't
> think this is possible without breaking old code and tests.
I was not suggesting that we keep both code paths. I was only thinking of being able to toggle a switch to run the same tests once using the Configuration-based and once using the new metamodel.
> It might be a good idea to have to checkouts. One with the current
> HEAD of the metamodel work
> and one with the old code still in place. Instead of doing everything
> on one checkout work in two
> different checkouts with two IDE windows etc.
I do not have enough memory on my laptop to be able to debug 2 Intellij projects at the same time. I suppose I could buy more memory and see how that goes.
> This is probably even
> more effective, since now you
> can really compare things side by side. I am not doing this yet,
> since in the current state of the metamodel
> the test targeting just the metamodel are independent of any settings
> (as Steve already pointed out).
I seem to be the only one writing tests that extend BaseCoreFunctionalTestCase with hibernate.test.new_metadata_mapping=true. I've been doing this to ensure that the metamodel info is translated into the persisters properly and that standard operations (i.e., SQL and bound parameters) are being executed properly.
Until this email thread, I was maintaining a separate copy of each test, one for each type of metadata. I only checked in the copy that uses the new metamodel. I deleted my copies that used the Configuration-based metadata, thinking I could use hibernate.test.new_metadata_mapping=false. I guess I should have pushed them to my fork first. I'll be re-creating them. Not a big deal, but kind of a pain.
TBH, I find it interesting that the people that are not actually doing this work think that simply being able to toggle a configuration parameter is a waste of time...
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
More information about the hibernate-dev