Response to a query. The main result is the nearest neighbor list, but we also optionally include debugging information if the appropriate field is set in the request. Last used: 8

Attributes

• latency (type: list(GoogleApi.ContentWarehouse.V1.Model.ResearchScamOnlineSearchLatencyStats.t), default: nil) - Debugging fields: The wall and CPU time used by each query on each machine. For successful queries, there will be one entry here for each machine that the query used if QueryRequest.enable_latency_stats was enabled. For unsuccessful queries, the contents of this field will be undefined. NOTE: The following accounting rules apply in batched mode, assuming there are
• numDeadlineExceededMachines (type: integer(), default: nil) - The number of non-root machines for which a DEADLINE_EXCEEDED error occurred when they were contacted. This value does NOT include machines for which no reply was received because their parent machine timed out. Therefore, num_total_machines may be larger than num_ok_machines + num_unreachable_machines + num_deadline_exceeded_machines.
• numOkMachines (type: integer(), default: nil) - Number of machines that contributed to the results; this might be less than the total number of machines if a machine has failed. The impact of a single machine failure could be larger than just one machine, if it happens to be one of the machines involved in distributing the query and collecting results.
• numTotalMachines (type: integer(), default: nil) - The number of machines used in this service for the dataset that was queried. If everything went right, this should be equal to num_ok_machines. If num_ok_machines < num_total_machines, some neighbors may be missing from results.
• numUnreachableMachines (type: integer(), default: nil) - The number of non-root machines for which an UNREACHABLE error occurred when they were contacted. This value does NOT include machines that were implicitly unreachable because their parent machine was not reachable. Therefore, num_total_machines may be larger than num_ok_machines + num_unreachable_machines + num_deadline_exceeded_machines.
• restrictStats (type: list(GoogleApi.ContentWarehouse.V1.Model.ResearchScamRestrictStats.t), default: nil) - The number of "active" datapoints for each dataset, i.e. points that are whitelisted by restricts and could be returned if they were close enough to the query. For successful queries, there will be one entry here for each dataset queried, if QueryRequest.enable_restrict_stats was true. For partially-successful successful queries, this field will contain results
• results (type: list(GoogleApi.ContentWarehouse.V1.Model.ResearchScamNearestNeighbors.t), default: nil) - The results for each dataset searched. If per_dataset_parameters was empty in QueryRequest then this will have one entry, the results for dataset 0. If per_dataset_parameters was not empty in QueryRequest, this will contain the results for each dataset queried, in order corresponding to the order of QueryRequest.per_dataset_parameters.
• status (type: GoogleApi.ContentWarehouse.V1.Model.UtilStatusProto.t, default: nil) - USEFUL ONLY IN CUSTOM BINARIES. In the stock ScaM server binary, all errors are relayed via the RPC's status. Thus, if RPC's status is ok, this field is guaranteed to be ok and if RPC's status is an error, no QueryResponse will be returned. The status of this query. This is useful when using a batched postprocessing lambda, which may ignore erroneous subqueries, continue postprocessing the valid ones and return OkStatus. In this case, this field will inform the client of any invalid subqueries.