[
https://issues.apache.org/jira/browse/OPENJPA-1371?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12772296#action_12772296
]
Fay Wang commented on OPENJPA-1371:
-----------------------------------
During merge, openjpa needs to retrieve id from the entity to decide whether
this is a new entity or a detached one. If it thinks it is a new entity, an
INSERT will be perform. Otherwise, an Update statement will be executed.
If (1) an entity has the compound primary key using IdClass, and
(2) some field in the IdClass is a derived identity from a
OneToOne/ManyToOne field,
(3) that value in that field has not yet become managed (i.e., does not
have the StateManagerImpl yet),
this field in the IdClass will not be populated, as the primary key of the
OneToOne/ManyToOne field usually can not be fetched back without an associated
StateManagerImpl, unless it has a single-value primary key.
Currently openjpa returns the null value for this field in the IdClass
regardlessly in the above situation. The patch provides a relief when this
derived identity-relation field has a single value primary key. It should fix
the problem reported by Constantine. However, the limitation for
nested/multi-level compound primary (e.g. Stock entity has compound primary key
using IdClass, and some value in the idClass is a derived-identity relation
field, ...) during merge remains.
> Insert is called instead of Update when merge() with derived Identity
> ---------------------------------------------------------------------
>
> Key: OPENJPA-1371
> URL: https://issues.apache.org/jira/browse/OPENJPA-1371
> Project: OpenJPA
> Issue Type: Bug
> Components: kernel
> Affects Versions: 2.0.0-M3
> Reporter: Fay Wang
> Fix For: 2.0.0-M3
>
>
> Insert is called instead of Update when merge() with derived Identity:
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
