mirror of
https://github.com/apache/druid.git
synced 2025-02-07 18:48:33 +00:00
4bfc186153
MSQ sorts the columns in a highly specialized manner by byte comparisons. As such the values are serialized differently. This works well for the primitive types and primitive arrays, however complex types cannot be serialized specially.
This PR adds the support for sorting the complex columns by deserializing the value from the field and comparing it via the type strategy. This is a lot slower than the byte comparisons, however, it's the only way to support sorting on complex columns that can have arbitrary serialization not optimized for MSQ.
The primitives and the arrays are still compared via the byte comparison, therefore this doesn't affect the performance of the queries supported before the patch. If there's a sorting key with mixed complex and primitive/primitive array types, for example: longCol1 ASC, longCol2 ASC, complexCol1 DESC, complexCol2 DESC, stringCol1 DESC, longCol3 DESC, longCol4 ASC, the comparison will happen like:
longCol1, longCol2 (ASC) - Compared together via byte-comparison, since both are byte comparable and need to be sorted in ascending order
complexCol1 (DESC) - Compared via deserialization, cannot be clubbed with any other field
complexCol2 (DESC) - Compared via deserialization, cannot be clubbed with any other field, even though the prior field was a complex column with the same order
stringCol1, longCol3 (DESC) - Compared together via byte-comparison, since both are byte comparable and need to be sorted in descending order
longCol4 (ASC) - Compared via byte-comparison, couldn't be coalesced with the previous fields as the direction was different
This way, we only deserialize the field wherever required
75 lines
7.8 KiB
Plaintext
75 lines
7.8 KiB
Plaintext
com.fasterxml.jackson.databind.ObjectMapper#reader(com.fasterxml.jackson.core.type.TypeReference) @ Use ObjectMapper#readerFor instead
|
|
com.fasterxml.jackson.databind.ObjectMapper#reader(com.fasterxml.jackson.databind.JavaType) @ Use ObjectMapper#readerFor instead
|
|
com.fasterxml.jackson.databind.ObjectMapper#reader(java.lang.Class) @ Use ObjectMapper#readerFor instead
|
|
com.fasterxml.jackson.databind.ObjectMapper#writeValue(com.fasterxml.jackson.core.JsonGenerator, java.lang.Object) @ Use JacksonUtils#writeObjectUsingSerializerProvider to allow SerializerProvider reuse
|
|
com.fasterxml.jackson.core.JsonGenerator#writeObject(java.lang.Object) @ Use JacksonUtils#writeObjectUsingSerializerProvider to allow SerializerProvider reuse
|
|
com.google.common.base.Charsets @ Use java.nio.charset.StandardCharsets instead
|
|
com.google.common.collect.Iterators#emptyIterator() @ Use java.util.Collections#emptyIterator()
|
|
com.google.common.collect.Iterators#mergeSorted(java.lang.Iterable,java.util.Comparator) @ Use org.apache.druid.java.util.common.collect.Utils#mergeSorted()
|
|
com.google.common.collect.Lists#newArrayList() @ Create java.util.ArrayList directly
|
|
com.google.common.collect.Lists#newLinkedList() @ Use ArrayList or ArrayDeque instead
|
|
com.google.common.collect.Lists#newLinkedList(java.lang.Iterable) @ Use ArrayList or ArrayDeque instead
|
|
com.google.common.collect.MapMaker @ Create java.util.concurrent.ConcurrentHashMap directly
|
|
com.google.common.collect.Maps#newConcurrentMap() @ Create java.util.concurrent.ConcurrentHashMap directly
|
|
com.google.common.collect.Maps#newHashMap() @ Create java.util.HashMap directly
|
|
com.google.common.collect.Maps#newHashMap(java.util.Map) @ Create java.util.HashMap directly
|
|
com.google.common.collect.Maps#newTreeMap() @ Create java.util.TreeMap directly
|
|
com.google.common.collect.Maps#newTreeMap(java.util.Comparator) @ Create java.util.TreeMap directly
|
|
com.google.common.collect.Maps#newTreeMap(java.util.SortedMap) @ Create java.util.TreeMap directly
|
|
com.google.common.collect.Queues#newArrayDeque() @ Create java.util.ArrayDeque directly
|
|
com.google.common.collect.Queues#newConcurrentLinkedQueue() @ Create java.util.concurrent.ConcurrentLinkedQueue directly
|
|
com.google.common.collect.Queues#newLinkedBlockingQueue() @ Create java.util.concurrent.LinkedBlockingQueue directly
|
|
com.google.common.collect.Sets#newHashSet() @ Create java.util.HashSet directly
|
|
com.google.common.collect.Sets#newLinkedHashSet() @ Create java.util.LinkedHashSet directly
|
|
com.google.common.collect.Sets#newTreeSet() @ Create java.util.TreeSet directly
|
|
com.google.common.collect.Sets#newTreeSet(java.util.Comparator) @ Create java.util.TreeSet directly
|
|
com.google.common.io.Files#createTempDir() @ Use org.apache.druid.java.util.common.FileUtils.createTempDir()
|
|
java.io.File#mkdirs() @ Use org.apache.druid.java.util.common.FileUtils.mkdirp instead
|
|
java.io.File#toURL() @ Use java.io.File#toURI() and java.net.URI#toURL() instead
|
|
java.lang.String#matches(java.lang.String) @ Use startsWith(), endsWith(), contains(), or compile and cache a Pattern explicitly
|
|
java.lang.String#replace(java.lang.CharSequence,java.lang.CharSequence) @ Use one of the appropriate methods in StringUtils instead
|
|
java.lang.String#replaceAll(java.lang.String,java.lang.String) @ Use one of the appropriate methods in StringUtils instead, or compile and cache a Pattern explicitly
|
|
java.lang.String#replaceFirst(java.lang.String,java.lang.String) @ Use String.indexOf() and substring methods, or compile and cache a Pattern explicitly
|
|
java.nio.file.Files#createTempDirectory(java.lang.String,java.nio.file.attribute.FileAttribute[]) @ Use org.apache.druid.java.util.common.FileUtils.createTempDir()
|
|
java.nio.file.Files#createTempDirectory(java.nio.file.Path,java.lang.String,java.nio.file.attribute.FileAttribute[]) @ Use org.apache.druid.java.util.common.FileUtils.createTempDir()
|
|
java.util.HashMap#<init>(int) @ Use com.google.common.collect.Maps#newHashMapWithExpectedSize(int) instead
|
|
java.util.HashMap#<init>(int, float) @ Use com.google.common.collect.Maps#newHashMapWithExpectedSize(int) instead
|
|
java.util.LinkedHashMap#<init>(int) @ Use org.apache.druid.utils.CollectionUtils#newLinkedHashMapWithExpectedSize(int) instead
|
|
java.util.LinkedHashMap#<init>(int, float) @ Use org.apache.druid.utils.CollectionUtils#newLinkedHashMapWithExpectedSize(int) instead
|
|
java.util.HashSet#<init>(int) @ Use com.google.collect.Sets#newHashSetWithExpectedSize(int) instead
|
|
java.util.HashSet#<init>(int, float) @ Use com.google.collect.Sets#newHashSetWithExpectedSize(int) instead
|
|
java.util.LinkedHashSet#<init>(int) @ Use com.google.collect.Sets#newLinkedHashSatWithExpectedSize(int) instead
|
|
java.util.LinkedHashSet#<init>(int, float) @ Use com.google.collect.Sets#newLinkedHashSatWithExpectedSize(int) instead
|
|
java.util.LinkedList @ Use ArrayList or ArrayDeque instead
|
|
java.util.Random#<init>() @ Use ThreadLocalRandom.current() or the constructor with a seed (the latter in tests only!)
|
|
java.lang.Math#random() @ Use ThreadLocalRandom.current()
|
|
java.util.regex.Pattern#matches(java.lang.String,java.lang.CharSequence) @ Use String.startsWith(), endsWith(), contains(), or compile and cache a Pattern explicitly
|
|
org.apache.calcite.sql.type.OperandTypes#LITERAL @ LITERAL type checker throws when literals with CAST are passed. Use org.apache.druid.sql.calcite.expression.DefaultOperandTypeChecker instead.
|
|
org.apache.calcite.sql.type.OperandTypes#BOOLEAN_LITERAL @ Create a type checker like org.apache.calcite.sql.type.POSITIVE_INTEGER_LITERAL and use that instead
|
|
org.apache.calcite.sql.type.OperandTypes#ARRAY_BOOLEAN_LITERAL @ Create a type checker like org.apache.calcite.sql.type.POSITIVE_INTEGER_LITERAL and use that instead
|
|
org.apache.calcite.sql.type.OperandTypes#POSITIVE_INTEGER_LITERAL @ Use org.apache.calcite.sql.type.POSITIVE_INTEGER_LITERAL instead
|
|
org.apache.calcite.sql.type.OperandTypes#UNIT_INTERVAL_NUMERIC_LITERAL @ Create a type checker like org.apache.calcite.sql.type.POSITIVE_INTEGER_LITERAL and use that instead
|
|
org.apache.calcite.sql.type.OperandTypes#NUMERIC_UNIT_INTERVAL_NUMERIC_LITERAL @ Create a type checker like org.apache.calcite.sql.type.POSITIVE_INTEGER_LITERAL and use that instead
|
|
org.apache.calcite.sql.type.OperandTypes#NULLABLE_LITERAL @ Create an instance of org.apache.calcite.sql.type.CastedLiteralOperandTypeChecker that allows nulls and use that instead
|
|
org.apache.commons.io.FileUtils#getTempDirectory() @ Use org.junit.rules.TemporaryFolder for tests instead
|
|
org.apache.commons.io.FileUtils#deleteDirectory(java.io.File) @ Use org.apache.druid.java.util.common.FileUtils#deleteDirectory()
|
|
org.apache.commons.io.FileUtils#forceMkdir(java.io.File) @ Use org.apache.druid.java.util.common.FileUtils.mkdirp instead
|
|
org.apache.datasketches.memory.Memory#wrap(byte[], int, int, java.nio.ByteOrder) @ The implementation isn't correct in datasketches-memory-2.2.0. Please refer to https://github.com/apache/datasketches-memory/issues/178. Use wrap(byte[]) and modify the offset by the callers instead
|
|
java.lang.Class#getCanonicalName() @ Class.getCanonicalName can return null for anonymous types, use Class.getName instead.
|
|
java.util.concurrent.Executors#newFixedThreadPool(int) @ Executor is non-daemon and can prevent JVM shutdown, use org.apache.druid.java.util.common.concurrent.Execs#multiThreaded(int, java.lang.String) instead.
|
|
|
|
@defaultMessage Use Locale.ENGLISH
|
|
com.ibm.icu.text.DateFormatSymbols#<init>()
|
|
com.ibm.icu.text.SimpleDateFormat#<init>()
|
|
com.ibm.icu.text.SimpleDateFormat#<init>(java.lang.String)
|
|
|
|
@defaultMessage For performance reasons, use the utf8Base64 / encodeBase64 / encodeBase64String / decodeBase64 / decodeBase64String methods in StringUtils
|
|
org.apache.commons.codec.binary.Base64
|
|
com.google.common.io.BaseEncoding#base64()
|
|
|
|
@defaultMessage Use com.google.errorprone.annotations.concurrent.GuardedBy
|
|
javax.annotation.concurrent.GuardedBy
|
|
com.amazonaws.annotation.GuardedBy
|
|
|
|
org.powermock.** @ Use Mockito instead of Powermock for compatibility with newer Java versions
|