[GitHub] [druid] gianm commented on a change in pull request #9488: Match GREATEST/LEAST function behavior to other DBs

[GitBox <gi...@apache.org>]

gianm commented on a change in pull request #9488: Match GREATEST/LEAST function behavior to other DBs
URL: https://github.com/apache/druid/pull/9488#discussion_r391092673
 
 

 ##########
 File path: core/src/main/java/org/apache/druid/math/expr/Function.java
 ##########
 @@ -976,6 +981,163 @@ protected ExprEval eval(double x, double y)
     }
   }
 
+  class GreatestFunc extends ReduceFunc
+  {
+    public static final String NAME = "greatest";
+
+    public GreatestFunc()
+    {
+      super(
+          Math::max,
+          Math::max,
+          BinaryOperator.maxBy(Comparator.naturalOrder())
+      );
+    }
+
+    @Override
+    public String name()
+    {
+      return NAME;
+    }
+  }
+
+  class LeastFunc extends ReduceFunc
+  {
+    public static final String NAME = "least";
+
+    public LeastFunc()
+    {
+      super(
+          Math::min,
+          Math::min,
+          BinaryOperator.minBy(Comparator.naturalOrder())
+      );
+    }
+
+    @Override
+    public String name()
+    {
+      return NAME;
+    }
+  }
+
+  abstract class ReduceFunc implements Function
+  {
+    private final DoubleBinaryOperator doubleReducer;
+    private final LongBinaryOperator longReducer;
+    private final BinaryOperator<String> stringReducer;
+
+    ReduceFunc(
+        DoubleBinaryOperator doubleReducer,
+        LongBinaryOperator longReducer,
+        BinaryOperator<String> stringReducer
+    )
+    {
+      this.doubleReducer = doubleReducer;
+      this.longReducer = longReducer;
+      this.stringReducer = stringReducer;
+    }
+
+    @Override
+    public void validateArguments(List<Expr> args)
+    {
+      // anything goes
+    }
+
+    @Override
+    public ExprEval apply(List<Expr> args, Expr.ObjectBinding bindings)
+    {
+      if (args.isEmpty()) {
+        return ExprEval.of(null);
+      }
+
+      ExprAnalysis exprAnalysis = analyzeExprs(args, bindings);
+      if (exprAnalysis == null) {
+        // The GREATEST/LEAST functions are not in the SQL standard, but most (e.g., MySQL, Oracle) return NULL if any
+        // are NULL. Others (e.g., Postgres) only return NULL if all are NULL, otherwise the NULLs are ignored.
 
 Review comment:
   Given it's not standard and we can do what we want, I would suggest going with the Postgres behavior for these reasons:
   
   - IMO the Postgres behavior is more likely to be useful: if I do `GREATEST(x, y, z)` and `y` is null, I probably want that to be equivalent to `GREATEST(x, z)`.
   - Postgres behavior is used as a base for a wide variety of other databases beyond the ones we're looking at here, so its behavior is influential.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org