[
https://issues.apache.org/jira/browse/DERBY-3856?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Knut Anders Hatlen updated DERBY-3856:
--------------------------------------
Attachment: fix.diff
Attaching a suggested fix for the issue. I have only run DateTimeTest on it yet.
Changes made by the patch:
java/engine/org/apache/derby/iapi/types/SQLTimestamp.java:
java/engine/org/apache/derby/iapi/types/SQLDate.java:
- Don't set valueString in parseTimestamp()/parseDate(). Since it's left as
null, it will be generated (and cached) on the first invocation of getString(),
and it will be on a normalized form.
java/testing/org/apache/derbyTesting/functionTests/tests/lang/DateTimeTest.java:
- Removed code to work around this issue.
- Added regression test cases for TIMESTAMP, DATE and TIME.
> difference between Embedded vs DerbyNetClient in format of return from
> timestamp(cast(? as varchar(32)))
> --------------------------------------------------------------------------------------------------------
>
> Key: DERBY-3856
> URL: https://issues.apache.org/jira/browse/DERBY-3856
> Project: Derby
> Issue Type: Bug
> Components: SQL
> Affects Versions: 10.3.3.1, 10.4.2.0, 10.5.1.1
> Reporter: Myrna van Lunteren
> Assignee: Knut Anders Hatlen
> Attachments: fix.diff
>
>
> There is a slight difference in how Embedded vs. DerbyNetClient return a
> specific cast.
> This showed up during conversion of the test lang/datetime.sql which before
> was only run with Embedded...
> The following sql:
> prepare dateTimePS as 'values( date(cast(? as integer)),timestamp(cast(? as
> varchar(32))))';
> execute dateTimePS using 'values(cast(1 as integer),
> ''2003-03-05-17.05.43.111111'')';
> gives:
> 1 |2
> -------------------------------------
> Embedded: 1970-01-01|2003-03-05-17.05.43.111111
> DerbyNetClient: 1970-01-01|2003-03-05 17:05:43.111111
> (in Embedded there's a '-' between date and time part, with DerbyNetClient a
> space; with Embedded the separator between time elements is ., with
> DerbyNetClient :. Embedded reflects the data as passed in, with
> DerbyNetClient it seems to be the default timestamp format).
> I am not sure which is correct at this point, but I confirmed the behavior is
> like this in latest builds of trunk and 10.3 and 10.4 branches.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
