[
https://issues.apache.org/jira/browse/PHOENIX-6003?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17680150#comment-17680150
]
Istvan Toth edited comment on PHOENIX-6003 at 1/24/23 9:12 AM:
---------------------------------------------------------------
Why is this still not fixed ?
The reason this is not closed is that if we encounter a combination where we
use a Phoenix version where PHOENIX-6493 is not fixed, but PQS has CALCITE-4138
, then we'd completely break HUE if
we set the proper has_ fields in the Python driver, as we'd correctly push
schema='' through Avatica, but then Phoenix couldn't handle it.
The current workaround in the Python driver works well enough with every
combination.
was (Author: stoty):
Why is this still not fixed ?
The reason this is not closed is that if we encounter a combinate where we use
a Phoenix version where PHOENIX-6493 is not fixed, but PQS has CALCITE-4138 ,
then we'd completely break HUE if
we set the proper has_ fields in the Python driver, as we correctly push
schema='' through Avatica, but then Phoenix couldn't handle it.
The current workaround in the Python driver works well enough with every
combination.
> Metadata operations via Avatica turn empty string args to null
> --------------------------------------------------------------
>
> Key: PHOENIX-6003
> URL: https://issues.apache.org/jira/browse/PHOENIX-6003
> Project: Phoenix
> Issue Type: Bug
> Components: queryserver
> Affects Versions: queryserver-6.0.0
> Reporter: Istvan Toth
> Priority: Major
>
> PhoenixDatabaseMetaData.getTables(), and some other functions have parameters
> (catalog, schemaPattern), where null and empty String have different
> semantics.
> The corresponding protobuf fields in Avatica are not nullable, and PQS seems
> to interpret empty string as Null, making it impossible to get the default
> catalog/schema only.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
