wuyue created ARROW-16088:
-----------------------------

             Summary: Possible gandiva function problem
                 Key: ARROW-16088
                 URL: https://issues.apache.org/jira/browse/ARROW-16088
             Project: Apache Arrow
          Issue Type: Bug
          Components: C++ - Gandiva
    Affects Versions: 7.0.0
         Environment: Aliyun ECS
CentOS Linux release 7.9.2009 (Core)
/arrow/cpp/src/gandiva/engine.cc:109: Detected CPU Name : skylake-avx512
openjdk:8u322-jdk
            Reporter: wuyue


I use Gandiva JNI  and function greater_than_or_equal_to is not support.

Is it really not support or i use it in a wrong way?
{code:java}
//代码占位符
List<Field> fields = new ArrayList<>();
Field idF = new Field("id", new FieldType(false, new ArrowType.Int(8, false), 
null), null);
fields.add(idF);
fields.add(new Field("name", new FieldType(false, new ArrowType.Utf8(), null), 
null));
Schema schema = new Schema(fields);

RootAllocator allocator = new RootAllocator(Long.MAX_VALUE);
VectorSchemaRoot root = VectorSchemaRoot.create(schema, allocator);
root.getFieldVectors().forEach(vec -> vec.setInitialCapacity(100));
root.allocateNew();

root.getFieldVectors().forEach(fv -> {
    if (fv instanceof IntVector) {
        for (int i = 0; i < 100; i++) {
            ((IntVector) fv).setSafe(i, i);
        }
    }
    if (fv instanceof VarCharVector) {
        for (int i = 0; i < 100; i++) {
            ((VarCharVector) fv).setSafe(i, String.valueOf(i + 100).getBytes());
        }
    }
});

List<TreeNode> list = new ArrayList<>();
TreeNode idNode = TreeBuilder.makeField(idF);
TreeNode literalNode = TreeBuilder.makeLiteral(50);
TreeNode eqNode = TreeBuilder.makeFunction("greater_than_or_equal_to", list, 
new ArrowType.Bool());
Condition condition = TreeBuilder.makeCondition(eqNode);
Filter filter = Filter.make(root.getSchema(), condition);

ArrowBuf selectionBuffer = allocator.buffer(200);
SelectionVectorInt16 selectionVector = new 
SelectionVectorInt16(selectionBuffer);

ArrowRecordBatch data = new VectorUnloader(root).getRecordBatch();
filter.evaluate(data, selectionVector); {code}
 



--
This message was sent by Atlassian Jira
(v8.20.1#820001)

Reply via email to