[
https://issues.apache.org/jira/browse/CALCITE-5771?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ZheHu updated CALCITE-5771:
---------------------------
Summary: Apply two different NULL semantics for CONCAT function(enabled in
MySQL、Postgresql、BigQuery and MSSQL) (was: Apply two different NULL semantics
for CONCAT function(enabled in MySQL、Postgresql and BigQuery))
> Apply two different NULL semantics for CONCAT function(enabled in
> MySQL、Postgresql、BigQuery and MSSQL)
> ------------------------------------------------------------------------------------------------------
>
> Key: CALCITE-5771
> URL: https://issues.apache.org/jira/browse/CALCITE-5771
> Project: Calcite
> Issue Type: Bug
> Components: core
> Affects Versions: 1.34.0
> Reporter: ZheHu
> Assignee: ZheHu
> Priority: Major
>
> We now have one CONCAT_FUNCTION which is enabled in
> MySQL、Postgresql、BigQuery, it always returns NULL when any of the arguments
> is NULL, and it accepts at least 2 arguments.
> The following table describes how different database products define CONCAT
> function:
> || DB Product || Argument type in CONCAT || Argument num in CONCAT || Result
> ||
> |MySQL|string|at least 1|returns null if any argument is null, otherwise
> return string|
> |Postgresql|any(cast to string)|at least 1|always return string, null is
> treated as empty string|
> |BigQuery|string|at least 1|returns null if any argument is null, otherwise
> return string|
> For MySQL and BigQuery, we can reuse the current CONCAT_FUNCTION(but need to
> change the accepted argument number). We also need another CONCAT function
> for Postgresql that treats NULL as empty String.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
