Norman Jordan created CALCITE-6550:
--------------------------------------
Summary: Improve SQL function overloading
Key: CALCITE-6550
URL: https://issues.apache.org/jira/browse/CALCITE-6550
Project: Calcite
Issue Type: Improvement
Components: core
Reporter: Norman Jordan
Assignee: Norman Jordan
All SQL functions across all SQL libraries as well as the standard functions
are used as keys in a Map (in the class {*}RexImpTable{*}). The hashcode and
equals methods currently only look at the name, kind and class.
[https://github.com/apache/calcite/blob/main/core/src/main/java/org/apache/calcite/sql/SqlOperator.java#L424]
[https://github.com/apache/calcite/blob/main/core/src/main/java/org/apache/calcite/sql/SqlOperator.java#L439]
This causes a few problems:
* Some functions may have the same name and kind but different sets of
parameters
* Two different SQL libraries may have functions with the same name and kind
(and even parameters) but have different implementations.
The current work around is to create new anonymous classes when needed to avoid
collisions in the Map. Here is an example.
https://github.com/apache/calcite/blob/main/core/src/main/java/org/apache/calcite/sql/fun/SqlLibraryOperators.java#L578
The hashcode and equals methods should consider more information, such as the
parameters as well as some way to differentiate similar functions in different
libraries.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
