Java 11 API Change Proposals

JDK11 features had been frozen since it entered the ramp-down phase last month. The big changes are listed as JEPs. In addition, there are many changes outside of the JEPs in JDK 11, but no summary exists. Therefore, I've listed the API changes in JDK 11 as far as I know.

String

lines()

Get a stream divided by line breaks:

jshell> "test\nhoge\n".lines().map(String::toUpperCase).toArray()
$11 ==> Object[2] { "TEST", "HOGE" }

repeat(int)

Repeat String for the specified number of times:

jshell> "test".repeat(3)
$7 ==> "testtesttest"

isBlank()

This method determines whether the string contains spaces. Full-width spaces are also treated as spaces.

jshell> var halfSpace = "\u0020"
halfSpace ==> " "

jshell> halfSpace.isBlank()
$11 ==> true

jshell> var fullSpace = "\u3000"
fullSpace ==> "　"

jshell> fullSpace.isBlank()
$13 ==> true

strip()/stripLeading()/stripTrailing()

While almost the same as  trim()/trimLeft()/trimRight(), this takes full-width spaces as a space.

jshell> var aaa = fullSpace + "aaa" + fullSpace
aaa ==> "　aaa　"

jshell> aaa.strip()
$14 ==> "aaa"

jshell> aaa.trim()
$15 ==> "　aaa　"

CharSequence

compare(CharSequence, CharSequence)

Sort by dictionary order.

It is used by compareTo() in CharSequence/StringBuffer/StringBuilder. Therefore, the three classes implement Comparable now.

Character

toString(int)

JDK 11 makes this process more convenient. 

JDK10.0.1

jshell> Character.toString(65)
|  Error:
|  incompatible types: possible lossy conversion from int to char
|  Character.toString(65)
|                     ^^

JDK11ea14

jshell> Character.toString(65)
$9 ==> "A"

Path

of(String, String...)

We had to use Paths.get(). Now, we can use of() in the same manner as other classes.

Files

writeString(Path, CharSequence)

We can save a String with one method.

jshell> Files.writeString(Path.of("test.txt"), "Hello!!!")
$3 ==> test.txt

readString(Path)

We can read a String with one method.

jshell> Files.readString(Path.of("test.txt"))
$4 ==> "Hello!!!"

Reader

nullReader()

We can get a  Reader  that does nothing.

Writer

nullWriter()

We can get a  Writer that does nothing.

InputStream

nullInputStream()

We can get an  InputStream  that does nothing.

OutputStream

nullOutputStream()

We can get an  OutputStream  that does nothing.

Predicate

not(Predicate)

We could choose to not use a method reference where it needs to invert a condition. Instead, now, we can use a method reference.

jshell> Stream.of("aa", "", "bb").filter(Predicate.not(String::isEmpty)).toArray()
$23 ==> Object[2] { "aa", "bb" }

Collection

toArray(IntFunction)

We had to use a non-stylish notation like list.toArray(new String[list.size())]) to create a typed array from a collection. Now, we can write in a stylish notation.

jshell> List.of("aa","bb").toArray(String[]::new)
$1 ==> String[2] { "aa", "bb" }

Optional/OptionalInt/OptionalLong/OptionalDouble

isEmpty()

 isPresent() had existed before now. Now, we haveisEmpty().

jshell> Optional.ofNullable(null).isEmpty()
$5 ==> true

TimeUnit

convert(Duration)

This has been added for java.util.concurrent.TimeUnit. 

Pattern

asMatchPredicate()

Up until now, there has only been asPredicate, but, now, we have asMatchPredicate.

jshell> var pred = Pattern.compile("aaa").asPredicate()
pred ==> java.util.regex.Pattern$$Lambda$25/0x00000008000b5040@2f686d1f

jshell> pred.test("aaa")
$6 ==> true

jshell> pred.test("aaab")
$7 ==> true

jshell> var matPred = Pattern.compile("aaa").asMatchPredicate()
matP ==> java.util.regex.Pattern$$Lambda$24/0x00000008000b6440@402a079c

jshell> matPred.test("aaa")
$9 ==> true

jshell> matPred.test("aaab")
$10 ==> false

ListSelectionModel

 getSelectedIndices() / getSelectedCount() have been added.

Thread

destroy()/stop(Throwable)

Removed  destroy() . stop() remains.

Policy

javax.security.auth.Policy has been removed.

ArrayIndexOutOfBoundsException

The exception message has changed:

JDK10.0.1

jshell> new int[]{}[0]
|  java.lang.ArrayIndexOutOfBoundsException thrown: 0
|        at (#8:1)

JDK11ea14

jshell> new int[]{}[0]
|  Exception java.lang.ArrayIndexOutOfBoundsException: Index 0 out of bounds for length 0
|        at (#4:1)

IndexOutOfBoundsException

In this change, the hyphens have been removed in this message:

JDK10.0.1

jshell> List.of().get(0)
|  java.lang.IndexOutOfBoundsException thrown: Index 0 out-of-bounds for length 0
|        at Preconditions.outOfBounds (Preconditions.java:64)
|        at Preconditions.outOfBoundsCheckIndex (Preconditions.java:70)
|        at Preconditions.checkIndex (Preconditions.java:248)
|        at Objects.checkIndex (Objects.java:372)
|        at ImmutableCollections$List0.get (ImmutableCollections.java:106)
|        at (#6:1)

JDK11ea14

jshell> List.of().get(0)
|  Exception java.lang.ArrayIndexOutOfBoundsException: Index 0 out of bounds for length 0
|        at ImmutableCollections$ListN.get (ImmutableCollections.java:411)
|        at (#3:1)

System

arraycopy

JDK10

jshell> System.arraycopy(new int[0],0,new double[0],0,0)
|  java.lang.ArrayStoreException thrown

JDK11ea19

jshell> System.arraycopy(new int[0], 0, new double[0], 0, 0)
|  Exception java.lang.ArrayStoreException: arraycopy: type mismatch: can not copy int[] into double[]

setProperty(String, String)

Before, changing java.home would cause some trouble. But, now, the issues have been resolved.

Supporting Japanese New Era

The Japanese Imperial Era is scheduled to change to the new era on May 1st, 2019.

It is introduced as a NewEra placeholder.

It will be updated in JDK 12.0.1 after the Japanese government announces it.

JDK10.0.1

jshell> JapaneseDate.of(2019, 5, 1)
$15 ==> Japanese Heisei 31-05-01

JDK11 ea18

jshell> JapaneseDate.of(2019, 5, 1)
$3 ==> Japanese NewEra 1-05-01

We have not been able to get May 1st Heisei 31 as our JapaneseDate.

JDK10.0.1

jshell> JapaneseDate.of(JapaneseEra.HEISEI, 31, 5, 1)
$14 ==> Japanese Heisei 31-05-01

JDK11 ea18

jshell> JapaneseDate.of(JapaneseEra.HEISEI, 31, 5, 1)
|  Exception java.time.DateTimeException: year, month, and day not valid for Era
|        at JapaneseDate.of (JapaneseDate.java:231)
|        at (#2:1)

Base64

Since ea20, the encoding has become faster by using AVX512, but I cannot confirm it on Windows.

Boolean

parseBoolean

It has become a little faster to remove the redundant null check, they have claimed.

JDK10

    public static boolean parseBoolean(String s) {
        return ((s != null) && s.equalsIgnoreCase("true"));
    }

JDK11

    public static boolean parseBoolean(String s) {
        return "true".equalsIgnoreCase(s);
    }

I haven't confirmed the performance difference yet.

TimSort

TimSort is the main algorithm that is used in  Array.sort() and Collection.sort() .

It had a bug where it threw an ArrayIndexOutOfBoundsException for some sequences, but that seems to be fixed. I haven't confirmed it yet.