[Bug]: No error is returned when searching an empty collection with a vector that has a different schema DataType

[ThreadDao]

Is there an existing issue for this?

• [X] I have searched the existing issues

Environment

- Milvus version: 2.4 and master
- Deployment mode(standalone or cluster):
- MQ type(rocksmq, pulsar or kafka):    
- SDK version(e.g. pymilvus v2.0.0rc2): 2.5.0rc31
- OS(Ubuntu or CentOS): 
- CPU/Memory: 
- GPU: 
- Others:

Current Behavior

1. create a collection with binary vector
2. index and load
3. search with float32 vector and no error is returned

• case:

  @pytest.mark.tags(CaseLabel.L2)
  def test_debug(self):
      """
      target: test insert binary with dim not match
      method: insert binary data dim not equal to schema
      expected: raise exception
      """
      c_name = cf.gen_unique_str(prefix)
      collection_w = self.init_collection_wrap(
          name=c_name, schema=default_binary_schema)
      print(collection_w.schema)
  
      _index = {"index_type": "BIN_IVF_FLAT", "metric_type": "JACCARD", "params": {"nlist": 128}}
      collection_w.create_index(ct.default_binary_vec_field_name, index_params=_index)
      collection_w.load()
  
      # search_vectors = cf.gen_binary_vectors(1, dim=512)[0]
      search_vectors = cf.gen_vectors(1, dim=128)
      print(len(search_vectors[0]))
      search_params = {"metric_type": "JACCARD", "params": {"nprobe": 32}}
      res1 = collection_w.search(data=search_vectors, anns_field=ct.default_binary_vec_field_name,
                                 param=search_params, limit=10)[0]

• log:

  <schema>: {'auto_id': False, 'description': '', 'fields': [{'name': 'int64', 'description': '', 'type': <DataType.INT64: 5>, 'is_primary': True, 'auto_id': False}, {'name': 'float', 'description': '', 'type': <DataType.FLOAT: 10>}, ......  (api_request.py:37)
  {'auto_id': False, 'description': '', 'fields': [{'name': 'int64', 'description': '', 'type': <DataType.INT64: 5>, 'is_primary': True, 'auto_id': False}, {'name': 'float', 'description': '', 'type': <DataType.FLOAT: 10>}, {'name': 'varchar', 'description': '', 'type': <DataType.VARCHAR: 21>, 'params': {'max_length': 65535}}, {'name': 'binary_vector', 'description': '', 'type': <DataType.BINARY_VECTOR: 100>, 'params': {'dim': 128}}], 'enable_dynamic_field': False}
  [2024-06-05 17:58:24,060 - DEBUG - ci_test]: (api_request)  : [Collection.create_index] args: ['binary_vector', {'index_type': 'BIN_IVF_FLAT', 'metric_type': 'JACCARD', 'params': {'nlist': 128}}, 1200], kwargs: {'index_name': ''} (api_request.py:62)
  [2024-06-05 17:58:24,626 - DEBUG - ci_test]: (api_response) : Status(code=0, message=)  (api_request.py:37)
  [2024-06-05 17:58:24,626 - DEBUG - ci_test]: (api_request)  : [Collection.load] args: [None, 1, 180], kwargs: {} (api_request.py:62)
  [2024-06-05 17:58:27,236 - DEBUG - ci_test]: (api_response) : None  (api_request.py:37)
  128
  [2024-06-05 17:58:27,238 - DEBUG - ci_test]: (api_request)  : [Collection.search] args: [[[0.14820588760615686, 0.08046226024204095, 0.12797510693312997, 0.010988964358071326, 0.12966189729205876, 0.08250214726508029, 0.0386905181070105, 0.14133018178279147, 0.11278505497518053, 0.14332169423440566, 0.010586310006486436, 0.0468956475141539, 0.0017581711433013439, 0.12107221448933901, 0......, kwargs: {} (api_request.py:62)
  [2024-06-05 17:58:27,472 - DEBUG - ci_test]: (api_response) : data: ['[]'] , cost: 0  (api_request.py:37)

Expected Behavior

No response

Steps To Reproduce

No response

Milvus Log

No response

Anything else?

No response

[ThreadDao]

@binbinlv Help to add relevant cases by pymilvus sdk. thanf you~

[smellthemoon]

@binbin12580 Help to add relevant cases by pymilvus sdk. thanf you~

The previous conclusion is that no need to do check on an empty collection.

[yanliang567]

@czs007 similar to #33637, do you have any plan to improve this in milvus 2.4.x? /assign @czs007 /unassign

[yanliang567]

@czs007 please let us know if no

[ThreadDao]

by design