[CALCITE-6325] Add LOG function (enabled in Mysql and Spark library) #3789
Conversation
| @@ -512,8 +512,7 @@ public enum BuiltInMethod { | |||
| SAFE_DIVIDE(SqlFunctions.class, "safeDivide", double.class, double.class), | |||
| SAFE_MULTIPLY(SqlFunctions.class, "safeMultiply", double.class, double.class), | |||
| SAFE_SUBTRACT(SqlFunctions.class, "safeSubtract", double.class, double.class), | |||
| LOG(SqlFunctions.class, "log", long.class, long.class), | |||
| LOG2(SqlFunctions.class, "log2", long.class), | |||
| LOG(SqlFunctions.class, "log", long.class, long.class, int.class), | |||
There was a problem hiding this comment.
Maybe this requires an enum instead of a numeric flag?
There was a problem hiding this comment.
@tanclary I'm sorry that it took so long to improve it. If you have time, please review this PR.
Co-authored-by: Tanner Clary <tannerclary@google.com>
|
|
|
||
| map.put(LN, new LogImplementor()); | ||
| map.put(LOG, new LogImplementor()); | ||
| map.put(LOG10, new LogImplementor()); | ||
|
|
||
| map.put(LOG_MYSQLSPARK, new LogMysqlSparkImplementor()); |
There was a problem hiding this comment.
The SQL function name that you are registering is LOG, which is the same as what is added on line 651. The equals and hashCode methods used for the key is from the class SqlOperator and will only use the name+kind.
The result is that you overwrite the map entry from line 651. One way around this is set LOG_MYSQLSPARK to a new anonymous inner subclass of SqlBasicFunction. You would need to increase visibility of the SqlBasicFunction constructor.
LOG_MYSQLSPARK = new SqlBasicFunction(...) {};
https://issues.apache.org/jira/browse/CALCITE-6325