[hibernate-issues] [Hibernate-JIRA] Commented: (HHH-6430) MappedAttribute should make use of HibernateTypeDescriptor

[ http://opensource.atlassian.com/projects/hibernate/browse/HHH-6430?page=c... ] Hardy Ferentschik commented on HHH-6430: ---------------------------------------- At the moment only the java type is preserved, but we need to distinguish between java type and explicitly specified types (eg via _@Type_). {noformat} hardy_: sebersole: one question regarding attributes. which "type" do you expect in an attribute binding? Atm I am using the java type, but I am wondering whether I should already try to resolve it to a HibernateType. Not sure exactly what is expected here and where the conversion should occur [4:31pm] sebersole: hardy_: in the attribute itself? or the binding? [4:32pm] sebersole: oh you said binding [4:33pm] sebersole: getting the hibernate type is not always feasible [4:33pm] sebersole: reveng case [4:33pm] sebersole: think ManyToOne [4:34pm] hardy_: in particular I am talking about AbstractAttributeBinding [4:34pm] sebersole: have you seen org.hibernate.metamodel.binding.HibernateTypeDescriptor? [4:34pm] sebersole: we need *something* like that [4:35pm] sebersole: i am open to suggestions [4:35pm] sebersole: you've played in the details there [4:36pm] sebersole: i think this stuff is stil largely the same from my orig proto [4:36pm] hardy_: it is [4:36pm] sebersole: obviously that had not been "evolved" [4:36pm] sebersole: what do you think? [4:36pm] hardy_: so right now AbstractAttributeBinding has a HibernateTypeDescriptor [4:36pm] sebersole: like i said, the only rub is that hibernate Type will not work [4:37pm] sebersole: unfortunately [4:37pm] hardy_: and w/ the old state code we called - hibernateTypeDescriptor.setTypeName( state.getTypeName() ); [4:37pm] sebersole: thats an explicit type [4:37pm] hardy_: hence my question - atm from the annotation side 'state.getTypeName()' will be the fqcn of the attribute [4:38pm] sebersole: i just added some code to do the guessing [4:38pm] hardy_: that i've seen [4:38pm] sebersole: ok [4:38pm] sebersole: well not the stuff gail dd [4:38pm] sebersole: org.hibernate.metamodel.source.hbm.BindingCreator#resolveTypeInformation [4:40pm] sebersole: the absolute ideal case would be to be able to resolve to the o.h.type.Type here [4:40pm] sebersole: but thats not possible unfortunately [4:42pm] sebersole: hardy_: we probably need some extra info there [4:42pm] sebersole: thats descriptor type name should be reserved for explicitly referenced type [4:43pm] sebersole: i assume annotations has @Type or somesuch [4:43pm] hardy_: yes there is @Type [4:43pm] sebersole: <property ... type="..."/> is the hbm equiv [4:43pm] hardy_: so we need some additional information whether the type is explicit or not [4:44pm] sebersole: well i think : [4:44pm] sebersole: 1) explicit type name [4:44pm] sebersole: 2) java type name [4:44pm] sebersole: i guess thats it [4:44pm] sebersole: well params [4:45pm] sebersole: just saying its important t distinguish (1) and (2) [4:45pm] hardy_: and why can we not use HibernateTypeDescriptor ? [4:45pm] sebersole: we can [4:46pm] hardy_: ahh, must have misunderstood something, thought we cannot [4:46pm] hardy_: so really when I process the annotations I should not just keep the say java type [4:46pm] hardy_: but I should create a HibernateTypeDescriptor [4:47pm] sebersole: the attribute creates it [4:47pm] sebersole: iirc [4:47pm] sebersole: you just need to populate it [4:47pm] hardy_: sure, but to populate it I need to have the information available [4:47pm] sebersole: is this making sense [4:47pm] hardy_: it does [4:47pm] sebersole: ok [4:48pm] sebersole: you do [4:48pm] sebersole: @Type - (1) explicit type name [4:48pm] sebersole: Memeber.getType() - (2) java type name [4:48pm] hardy_: sure [4:48pm] sebersole: if no @Type, then (1) is nul [4:49pm] sebersole: and the stuff that uses this info knows how to interpret that [4:49pm] hardy_: but in the annotation code I create a so called SimpleAttribute instance which collects all sorts of information about the attribute [4:49pm] sebersole: the binding you mean? [4:50pm] sebersole: oh [4:50pm] sebersole: yeah i saw that i believe [4:50pm] hardy_: and then when I create the AttributeBinding for the metamodel I use the SimpleAttribute to set everything which needs setting [4:50pm] sebersole: sure [4:51pm] hardy_: so yes, the information I need is in the annotations, but I need to make sure I have it SimpleAttribute [4:51pm] sebersole: sure [4:51pm] sebersole: another thought [4:51pm] hardy_: I might even just add a HibernateTypeDescriptor to SimpleAttribute [4:51pm] sebersole: we might also want to make toOne avail as a flag [4:52pm] sebersole: so we know its "nature" [4:52pm] hardy_: but that's a tangent now to the type stuff, right? [4:52pm] sebersole: no [4:52pm] hardy_: toOne [4:52pm] sebersole: talking about HibernateTypeDescriptor [4:52pm] hardy_: i get it [4:52pm] sebersole: HibernateTypeDescriptor.toOne {noformat} -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira