[
https://issues.apache.org/jira/browse/DERBY-277?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Knut Anders Hatlen updated DERBY-277:
-------------------------------------
Attachment: d277.sql
Repro attached, based on the SQL statements above.
On 10.1, the query has these (incorrect) results:
1
|2
--------------------------------------------------------------------------------------------------------------------------------------------
102_fieldname
|1
103_fieldname
|1
2 rows selected
On 10.2 and later, it has these (correct) results:
1
|2
--------------------------------------------------------------------------------------------------------------------------------------------
3_fieldname
|1
4_fieldname
|1
2 rows selected
So it seems like the bug has been fixed now.
> Group By and TRIM do not cooperate with more than 1 'not like' statement
> ------------------------------------------------------------------------
>
> Key: DERBY-277
> URL: https://issues.apache.org/jira/browse/DERBY-277
> Project: Derby
> Issue Type: Bug
> Components: SQL
> Environment: java version "1.3.1_13"
> Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.1_13-b03)
> Java HotSpot(TM) Client VM (build 1.3.1_13-b03, mixed mode)
> Running on Windows XP
> Reporter: Peter Haighton
> Priority: Minor
> Attachments: d277.sql
>
>
> Performing a select that appends BIGINT and VARCHARs together returns some
> strange results when mixed in with more than 1 not 'not like' statement.
> An example is as follows:
> SELECT RTRIM(CHAR(na.id))||'_'||fieldname, count(*) FROM na, nae, s WHERE
> na.type=3 and nae.naid=na.id AND fieldvalue='checked' and s.id=nae.sid and
> s.emailaddress not like '[EMAIL PROTECTED]' group by na.id,fieldname
> na.id is a primary key BIG INT
> fieldname is a varchar
> returns something like
> 32_challenge_100_0 38
> 32_challenge_100_1 42
> 32_challenge_100_2 38
> 32_challenge_100_3 42
> 32_challenge_100_6 1
> 32_challenge_101_0 25
> 32_challenge_101_1 35
> 32_challenge_101_2 30
> 32_challenge_101_3 18
> but adding an extra "and userid not like" statement in such as
> SELECT RTRIM(CHAR(na.id))||'_'||fieldname, count(*) FROM na, nae, s WHERE
> na.type=3 and nae.naid=na.id AND fieldvalue='checked' and s.id=nae.sid and
> s.emailaddress not like '[EMAIL PROTECTED]' and s.emailaddress not like
> '%def.com' group by na.id,fieldname
> I get:
> 32645_challenge_100_0 38
> 32645_challenge_100_1 42
> 33399_challenge_100_2 38
> 33399_challenge_100_3 42
> 18199_challenge_100_6 1
> 32645_challenge_101_0 25
> 32645_challenge_101_1 35
> 32645_challenge_101_2 30
> 32645_challenge_101_3 18
> ...
> Notice the number 32 has been changed to a somewhat appearing random number.
> The problem only occurs when I add the trim at the beginning
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
