[ http://opensource.atlassian.com/projects/hibernate/browse/HHH-4833?page=c... ] Steve Ebersole commented on HHH-4833: ------------------------------------- Per your request: {code} [2010-01-22 12:56] <epbernard> sebersole: on HEM attribute factory. PluralAttributeMetadataImpl and SingularAttributeMetadataImpl do no honor targetElement (like in @OneToMany(targetElement=MyEntity.class) ) [2010-01-22 12:56] <epbernard> you basically get the type of the Java type system [2010-01-22 12:56] <sebersole> epbernard, for associations ok thats wrong [2010-01-22 12:56] <epbernard> Is there a reason why you did not take it from the Hibernate metamodel? [2010-01-22 12:56] <sebersole> what we had before was wrong though for value types [2010-01-22 12:57] <sebersole> yes ^^ [2010-01-22 12:57] <epbernard> ah ok [2010-01-22 12:57] <sebersole> Dates [2010-01-22 12:57] <sebersole> remeber [2010-01-22 12:57] <epbernard> vaguely but yes [2010-01-22 12:57] <sebersole> pretty sure we talked about that [2010-01-22 12:57] <sebersole> ok [2010-01-22 12:57] <epbernard> I am wondering [2010-01-22 12:57] <epbernard> if the user does use raw types for it's model [2010-01-22 12:57] <sebersole> so the metamodel (Type) always reports java.util.Date [2010-01-22 12:58] <epbernard> should the metamodel try and correct that? [2010-01-22 12:58] <epbernard> I mean replace with targetElement [2010-01-22 12:58] <epbernard> I will make sure it does not break at least [2010-01-22 12:58] <epbernard> but I am wondering if that's something we want to support actually [2010-01-22 12:58] <sebersole> i dont grok all the implications of that [2010-01-22 12:58] <sebersole> support which? [2010-01-22 12:59] <epbernard> Should the metamodel reflects the targetElement [2010-01-22 12:59] <epbernard> rather than what the Java model exposes [2010-01-22 13:00] <epbernard> in practice, the type-safe criteria won't be useable witht he static metamodel [2010-01-22 13:00] <sebersole> thats what i was saying i dont grok all the implications of [2010-01-22 13:00] <epbernard> but OTOH this user clearly does not care about type safety :) [2010-01-22 13:01] <sebersole> well the exposed type could be an interface right? [2010-01-22 13:01] <sebersole> and the targetEntity names the concrete clas [2010-01-22 13:01] <sebersole> really this is the type of stuff i was getting at in the split i suggested in the new metamodel [2010-01-22 13:02] <sebersole> again not sure of names here yet as its still a work in progress... [2010-01-22 13:03] <sebersole> but the "logical model" would reflect the targetEntity relations [2010-01-22 13:03] <sebersole> but the "java model" or "object model" would reflect the declared types [2010-01-22 13:04] <sebersole> in the short term, i think the static metamodel should reflect the declared types [2010-01-22 13:04] <sebersole> iiuc [2010-01-22 13:06] <epbernard> sebersole: well that's clearly the easiest to implement so let's go for it :) [2010-01-22 13:07] <sebersole> epbernard, if you think its more techincally correct for the other... [2010-01-22 13:08] <sebersole> i think its actually pretty easy in terms of populating it (no clue about the "generating it" side) [2010-01-22 13:08] <sebersole> and could be pretty isolated to AttributeFactory [2010-01-22 13:08] <epbernard> sebersole: I've half done the generation actually [2010-01-22 13:08] <epbernard> it's not too hard [2010-01-22 13:09] <sebersole> if the metamodel type is an entity type (Type.isEntity()), prefer its targetType [2010-01-22 13:09] <sebersole> else, prefer the discovered (declared) type {code} -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HHH-4833?page=c... ] Emmanuel Bernard commented on HHH-4833: --------------------------------------- Intuitively I think it should honor targetEntity, that's why I've started the work on METAGEN-20. But that's not a big priority. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira