openjpa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Donald Woods (JIRA)" <>
Subject [jira] Updated: (OPENJPA-1414) Performance improvements for detach
Date Wed, 24 Feb 2010 14:36:28 GMT


Donald Woods updated OPENJPA-1414:

    Fix Version/s:     (was: 2.0.0-beta2)

> Performance improvements for detach
> -----------------------------------
>                 Key: OPENJPA-1414
>                 URL:
>             Project: OpenJPA
>          Issue Type: Improvement
>          Components: kernel
>    Affects Versions: 1.2.1
>            Reporter: Dianne Richards
>            Assignee: Dianne Richards
>             Fix For: 1.2.2, 1.3.0, 2.0.0-beta
>         Attachments: 1.2.x_doc_patch.txt, 2.0_doc_patch.txt, DM_1.2.x_patch.txt, DM_2.0_patch.txt,
> In performance runs, a lot of the time is spent on detach processing. This is partially
because a detach all is always done after an EntityManager clear() or close(). Therefore,
we nee to make the detach processing as efficient as possible. There are 2 areas that we've
found where this can be done:
> 1 -  DetachManager always attempts to load fields when detaching all. This does not appear
to be necessary when the default DetachState option of "loaded" is used.
> 2 - Detach goes through a bunch of processing to "detach version", which essentially
copies the version to the entity that is to be returned. But, when detach all internal is,
no copy is being made of the original entity. And, it appears that the original entity always
has the version. So, this processing is not necessary in this scenario.
> To be cautious, a new Compatibility property option is being created - reloadOnDetach
- to control both of these behavior. In 1.2.x, it will be true by default, so behavior will
not be changed for current users. The user will have to set this property to false in order
to get the performance benefit.
> If all goes well, the property will be set to false by default in 2.0.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message