// as a List() call is an important performance improvement.
		//
		// Spark based s3a committers are a  big enough use-case to have this optimization.
		//
		// A sample code to see the improvements is as follows, this sample code is
		// simply a read on JSON from MinIO and write it back as "parquet".
		//
		// import org.apache.spark.sql.SparkSession

			wantResult: `{"_1":3}`,
		},
		{
			name:       "date_diff_month_neg",
			query:      `SELECT date_diff(MONTH, '2020-01-20T', '2019-10-20T') FROM S3Object LIMIT 1`,
			wantResult: `{"_1":-3}`,
		},
		// Examples from https://docs.aws.amazon.com/AmazonS3/latest/userguide/s3-glacier-select-sql-reference-date.html#s3-glacier-select-sql-reference-date-diff
		{
			name:       "date_diff_year",

		{
			[]CompletePart{
				{ETag: validPartMD5, PartNumber: 6},
			},
		},
	}
	s3MD5 := getCompleteMultipartMD5(inputParts[3].parts)

	// Test cases with sample input values for CompleteMultipartUpload.
	testCases := []struct {
		bucket   string
		object   string
		uploadID string
		parts    []CompletePart
		// Expected output of CompleteMultipartUpload.

// Usage hint: Should be used at the end of the API end points tests (ex: check the last few lines of `testAPIListObjectPartsHandler`),
// need a sample HTTP request to be sent as argument so that the relevant handler is called, the handler registration is expected
// to be done since its called from within the API handler tests, the reference to the registered HTTP handler has to be sent

mc event list myminio/images
arn:minio:sqs::1:nats s3:ObjectCreated:*,s3:ObjectRemoved:* Filter: suffix=”.jpg”
```

### Step 3: Test on NATS

If you use NATS server, check out this sample program below to log the bucket notification added to NATS.

```go
package main

// Import Go and NATS packages
import (
 "log"
 "runtime"

 "github.com/nats-io/nats.go"
)

		// Multiple values for the same key (one map entry, longer slice) are cheaper
		// than the same number of values for different keys (many map entries), but
		// using a consistent per-value cost for overhead is simpler.
		maxMemoryBytes := 2 * int64(10<<20)
		maxMemoryBytes -= int64(len(name))
		maxMemoryBytes -= mapEntryOverhead
		if maxMemoryBytes < 0 {
			// We can't actually take this path, since nextPart would already have

	diskPath := tb.TempDir()

	// Initialize a new xlStorage layer.
	storage, err := newLocalXLStorageWithDiskIdx(diskPath, 3)
	if err != nil {
		return nil, "", err
	}

	// Create a sample format.json file

			expectedValidationErr: errLifecycleDuplicateID,
		},
		// Missing <Tag> in <And>
		{
			inputConfig:           `<LifecycleConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/"><Rule><ID>sample-rule-2</ID><Filter><And><Prefix>/a/b/c</Prefix></And></Filter><Status>Enabled</Status><Expiration><Days>1</Days></Expiration></Rule></LifecycleConfiguration>`,
			expectedParsingErr:    nil,
			expectedValidationErr: errXMLNotWellFormed,

			}
			opts.claims[k] = v
		}
	} else if globalIAMSys.LDAPConfig.Enabled() {
		// In case of LDAP we need to resolve the targetUser to a DN and
		// query their groups:
		opts.claims[ldapUserN] = targetUser // simple username
		targetUser, targetGroups, err = globalIAMSys.LDAPConfig.LookupUserDN(targetUser)
		if err != nil {
			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
			return
		}