You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by "Myrna van Lunteren (Commented) (JIRA)" <ji...@apache.org> on 2012/04/11 01:07:18 UTC

[jira] [Commented] (DERBY-5687) Back out the concurrency improvements for identity columns introduced by derby-4437

    [ https://issues.apache.org/jira/browse/DERBY-5687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13251158#comment-13251158 ] 

Myrna van Lunteren commented on DERBY-5687:
-------------------------------------------

For what it's worth, with sane jars built sync-ed up to revision 1311499 I ran the 'Embedded small' nstest 10 times on a windows machine where before I would see at least 3 to 5 of the runs have SYSCOLUMNS deadlocks, and/or Asserts complaining about either transaction not pristine, or transaction changed on live cacheable...And this time there were none of these problems.

I'll start closing the various bugs that represent these issues.
                
> Back out the concurrency improvements for identity columns introduced by derby-4437
> -----------------------------------------------------------------------------------
>
>                 Key: DERBY-5687
>                 URL: https://issues.apache.org/jira/browse/DERBY-5687
>             Project: Derby
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 10.9.0.0
>            Reporter: Rick Hillegas
>         Attachments: derby-5687-01-aa-backOutImprovement.diff, derby-5687-02-aa-publicAPI.diff, derby-5687-03-aa-adjustUserDocs.diff, derby-5687-03-aa-adjustUserDocs.diff
>
>
> DERBY-4437 attempted to improve the concurrency of identity columns by using SYSSEQUENCE-style sequence generators. These improvements caused NsTest to behave differently than it used to and they disclosed a problem in clearing the identity cache. The community lost confidence in this solution and it was backed out of the 10.8 branch under issue DERBY-5448. This new issue is filed to back the improvements out of the 10.9 trunk. Further useful discussion about how to improve the concurrency and correctness of identity columns has been taking place on DERBY-5443 and DERBY-5493.
> For the 10.9 release, identity columns will return to their old behavior of being ill-suited for high concurrency applications. Applications which need higher concurrency should be re-coded to use sequences rather than identity columns.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira