[
https://issues.apache.org/jira/browse/DERBY-6903?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16033787#comment-16033787
]
ASF subversion and git services commented on DERBY-6903:
--------------------------------------------------------
Commit 1797283 from [~bryanpendleton] in branch 'code/trunk'
[ https://svn.apache.org/r1797283 ]
DERBY-6903: ALTER TABLE ALTER COLUMN resets CYCLE option
DERBY-6904: Enhance ALTER TABLE to allow CYCLE option to be changed
DERBY-6905: Enhance SYS.SYSCOLUMNS to report the CYCLE value
DERBY-6906: Allow a GENERATED column to explicitly specify NO CYCLE
This patch was contributed by Harshvardhan Gupta (harshvardhan145 at gmail dot
com)
The patch adds the CYCLE property as a fundamental property associated
with a column similar to current increment value, current increment step etc.
* SYS.SYSCOLUMNS extended to hold the cycle value of identity column.
* Extended the grammar to accept "cycle","no cycle" in alter clause
and changed the underlying column descriptor and sequencing objects.
* ALTER clause not involving cycle option (i.e increment and restart)
won't affect the cycle option.
The code flow for the upgrade handling was modelled after Derby-534
for introducing a new in column SYSTRIGGERS. There are new behaviors
for SYSCOLUMNS in both soft- and hard-upgrade scenarios.
> ALTER TABLE ALTER COLUMN resets CYCLE option of IDENTITY column
> ---------------------------------------------------------------
>
> Key: DERBY-6903
> URL: https://issues.apache.org/jira/browse/DERBY-6903
> Project: Derby
> Issue Type: Sub-task
> Components: SQL
> Affects Versions: 10.13.1.0
> Reporter: Bryan Pendleton
> Assignee: Harshvardhan Gupta
> Priority: Minor
>
> Using ALTER TABLE with the SET INCREMENT BY or SET RESTART WITH
> options to make a change to a GENERATED ... BY IDENTITY column resets the
> CYCLE option of that column to "off".
> Some test cases demonstrating this behavior were added as part of
> the DERBY-6852 change revision 1756287
> The problem occurs when the ALTER TABLE code drops and recreates
> the underlying SEQUENCE object, and does not preserve the CYCLE
> option of the underlying SEQUENCE object.
> I believe the correct behavior would be for the ALTER TABLE code
> to read the current CYCLE option of the underlying SEQUENCE object
> first, and then create the new SEQUENCE object with the same value
> for the CYCLE option.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)