[JBoss JIRA] (TEIID-4455) Impala Translator - Change planning step for from_unixtime() pushdown
by Steven Hawkins (JIRA)
[ https://issues.jboss.org/browse/TEIID-4455?page=com.atlassian.jira.plugin... ]
Steven Hawkins commented on TEIID-4455:
---------------------------------------
> So is it expected that when user invokes from_unixtime against teiid instance, it may serve different results based on what source is this pushed against?
Ideally the driver will make the connection timezone the same as the Teiid server timezone. However if that is not occurring then yes you could get two different values.
> Impala Translator - Change planning step for from_unixtime() pushdown
> ---------------------------------------------------------------------
>
> Key: TEIID-4455
> URL: https://issues.jboss.org/browse/TEIID-4455
> Project: Teiid
> Issue Type: Feature Request
> Components: JDBC Connector
> Reporter: Don Krapohl
> Assignee: Kylin Soong
> Priority: Minor
> Labels: Impala_Translator
> Fix For: 9.2, 8.12.15.6_4
>
>
> Impala and Teiid both have from_unixtime() functions and per the forum reference the mapping to timestampadd() appears to come before the test for pushdown support of the impala native function.
> As a query request we require the impala from_unixtime() function be executed instead of the Teiid-native version.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months
[JBoss JIRA] (TEIID-4455) Impala Translator - Change planning step for from_unixtime() pushdown
by Jan Stastny (JIRA)
[ https://issues.jboss.org/browse/TEIID-4455?page=com.atlassian.jira.plugin... ]
Jan Stastny commented on TEIID-4455:
------------------------------------
> By default the values are considered to be in UTC timezone
Sorry, that's not true for Hive. It returns local timezone based timestamp.
So is it expected that when user invokes from_unixtime against teiid instance, it may serve different results based on what source is this pushed against?
> Impala Translator - Change planning step for from_unixtime() pushdown
> ---------------------------------------------------------------------
>
> Key: TEIID-4455
> URL: https://issues.jboss.org/browse/TEIID-4455
> Project: Teiid
> Issue Type: Feature Request
> Components: JDBC Connector
> Reporter: Don Krapohl
> Assignee: Kylin Soong
> Priority: Minor
> Labels: Impala_Translator
> Fix For: 9.2, 8.12.15.6_4
>
>
> Impala and Teiid both have from_unixtime() functions and per the forum reference the mapping to timestampadd() appears to come before the test for pushdown support of the impala native function.
> As a query request we require the impala from_unixtime() function be executed instead of the Teiid-native version.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months
[JBoss JIRA] (TEIID-5457) Issues with sybase/ms sql pad pushdown
by Steven Hawkins (JIRA)
[ https://issues.jboss.org/browse/TEIID-5457?page=com.atlassian.jira.plugin... ]
Steven Hawkins resolved TEIID-5457.
-----------------------------------
Fix Version/s: 10.3.4
11.0.2
Resolution: Done
Corrected the pushdown to account for null values and use the left function to appropriately trim when the pad length is shorter than the string.
> Issues with sybase/ms sql pad pushdown
> --------------------------------------
>
> Key: TEIID-5457
> URL: https://issues.jboss.org/browse/TEIID-5457
> Project: Teiid
> Issue Type: Bug
> Components: JDBC Connector
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Fix For: 11.1, 10.3.4, 11.0.2
>
>
> Sybase will return a non-null result then padding a null string.
> Sybase and ms sql will trim from the string from the left when lpad is specified with a pad length shorter than the string. The expected behavior is to always trim from the right.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months
[JBoss JIRA] (TEIID-4455) Impala Translator - Change planning step for from_unixtime() pushdown
by Steven Hawkins (JIRA)
[ https://issues.jboss.org/browse/TEIID-4455?page=com.atlassian.jira.plugin... ]
Steven Hawkins commented on TEIID-4455:
---------------------------------------
The return type as string was later addressed - TEIID-5026
> Teiid does not account for the timezone difference between source and running instance (ignores DatabaseTimeZone property). It is serving exactly the same timestamp as received from hive/impala instance, even though there's timezone difference.
The calendar based methods (TEIID-3847) were are not supported by the driver, so the property has no affect. It appears that is still the case for Hive - https://github.com/apache/hive/blob/master/jdbc/src/java/org/apache/hive/...
> Impala Translator - Change planning step for from_unixtime() pushdown
> ---------------------------------------------------------------------
>
> Key: TEIID-4455
> URL: https://issues.jboss.org/browse/TEIID-4455
> Project: Teiid
> Issue Type: Feature Request
> Components: JDBC Connector
> Reporter: Don Krapohl
> Assignee: Kylin Soong
> Priority: Minor
> Labels: Impala_Translator
> Fix For: 9.2, 8.12.15.6_4
>
>
> Impala and Teiid both have from_unixtime() functions and per the forum reference the mapping to timestampadd() appears to come before the test for pushdown support of the impala native function.
> As a query request we require the impala from_unixtime() function be executed instead of the Teiid-native version.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months
[JBoss JIRA] (TEIID-4455) Impala Translator - Change planning step for from_unixtime() pushdown
by Jan Stastny (JIRA)
[ https://issues.jboss.org/browse/TEIID-4455?page=com.atlassian.jira.plugin... ]
Jan Stastny commented on TEIID-4455:
------------------------------------
[~shawkins] I believe that direct pushdown is incorrect.
On impala and hive the return type of from_unixtime is string, which is not conveying the timezone information. (By default the values are considered to be in UTC timezone, but this can be tweaked at least in impala by a property '-use_local_tz_for_unix_timestamp_conversions' when starting the instance.)
Teiid does not account for the timezone difference between source and running instance (ignores DatabaseTimeZone property). It is serving exactly the same timestamp as received from hive/impala instance, even though there's timezone difference.
> Impala Translator - Change planning step for from_unixtime() pushdown
> ---------------------------------------------------------------------
>
> Key: TEIID-4455
> URL: https://issues.jboss.org/browse/TEIID-4455
> Project: Teiid
> Issue Type: Feature Request
> Components: JDBC Connector
> Reporter: Don Krapohl
> Assignee: Kylin Soong
> Priority: Minor
> Labels: Impala_Translator
> Fix For: 9.2, 8.12.15.6_4
>
>
> Impala and Teiid both have from_unixtime() functions and per the forum reference the mapping to timestampadd() appears to come before the test for pushdown support of the impala native function.
> As a query request we require the impala from_unixtime() function be executed instead of the Teiid-native version.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months
[JBoss JIRA] (TEIID-5458) Allow pi to convert from object/variant type
by Steven Hawkins (JIRA)
Steven Hawkins created TEIID-5458:
-------------------------------------
Summary: Allow pi to convert from object/variant type
Key: TEIID-5458
URL: https://issues.jboss.org/browse/TEIID-5458
Project: Teiid
Issue Type: Enhancement
Components: JDBC Connector
Reporter: Steven Hawkins
Assignee: Steven Hawkins
Fix For: 11.1, 11.0.2
Pi supports a variant type that allows for conversion to other types. This should be reflected by the translator.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months
[JBoss JIRA] (TEIID-5457) Issues with sybase/ms sql pad pushdown
by Steven Hawkins (JIRA)
[ https://issues.jboss.org/browse/TEIID-5457?page=com.atlassian.jira.plugin... ]
Steven Hawkins updated TEIID-5457:
----------------------------------
Description:
Sybase will return a non-null result then padding a null string.
Sybase and ms sql will trim from the string from the left when lpad is specified with a pad length shorter than the string. The expected behavior is to always trim from the right.
was:
Sybase will return a non-null result then padding a null string.
Sybase and ms sql will trim from the string from the left when lpad is specified with a pad length shorter than the string. The expected behavior is to always trim from the right.
The length of the total string is also not computed correctly.
> Issues with sybase/ms sql pad pushdown
> --------------------------------------
>
> Key: TEIID-5457
> URL: https://issues.jboss.org/browse/TEIID-5457
> Project: Teiid
> Issue Type: Bug
> Components: JDBC Connector
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Fix For: 11.1
>
>
> Sybase will return a non-null result then padding a null string.
> Sybase and ms sql will trim from the string from the left when lpad is specified with a pad length shorter than the string. The expected behavior is to always trim from the right.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
6 years, 4 months