Now that Java 17 is features complete (Rampdown Phase Two at the day of writing), it\u2019s time to walk throught all the functionalities that brings to us, developers, this new version.<\/p>\n
This article is part of a series on what\u2019s new on the last versions of Java<\/a>, for those who wants to read the others, here are the links : Java 16<\/a>, Java 15<\/a>, Java 14<\/a>, Java 13<\/a>, Java 12<\/a>, Java 11<\/a>,\u00a0Java 10,<\/a>\u00a0and\u00a0Java 9<\/a>.<\/p>\n This release doesn’t contain many JEPs, and therefore not many important new features. On the other hand, it brings some nice additions, and its LTS (Long Term Support) status makes it an important release.<\/p>\n This is probably the biggest news of this release, the pattern matching arrives in the switch, in preview. We can now make a switch on the type of a variable (including enum, record and array), and extract a local variable of the corresponding type.<\/p>\n The JEP gives the following example with an if\/else chain:<\/p>\n Which can now be re-written with a switch expression:<\/p>\n In addition to supporting pattern matching, the switch now allows you to define a special If we take the previous example, this gives us:<\/p>\n The big advantage is that it avoids having to do a defensive test before the switch, and allows to include in it the null value as any other possible values of our variable.<\/p>\n And it doesn’t stop there, the switch has been enhanced with guards<\/em> that allow to include a condition to the More information in the JEP-406<\/a>.<\/p>\n The JEP-356 provides a new interface New random number generation algorithms have been implemented, more secure and more powerful (but they are not thread-safe anymore), they are intended to replace the old ones.<\/p>\n Here are some examples of instantiation of random number generators:<\/p>\n More information in the JEP-356<\/a> or in this very complete article of\nMichael Bien : Java 17’s Enhanced Pseudo-Random Number Generators<\/a>.<\/p>\n In Java 17, only one feature goes from preview to standard: the Sealed Classes (JEP-409<\/a>), you can refer to my previous articles for more information.<\/p>\n The following features remain in preview (or incubator module).<\/p>\n For details on these, you can refer to my previous articles.<\/p>\n Java 17 adds support for the macOS\/AArch64 (aka Apple Silicon M1) architecture. More information about this in the JEP-391<\/a>.<\/p>\n The class More information in the ticket JDK-8251989<\/a>.<\/p>\n Testing code containing date manipulation has always been a challenge, especially if it uses and abuses To facilitate the testability of this kind of code, a new interface has been added to the JDK: Imagine the following code:<\/p>\n In normal operation, the Various JDK additions:<\/p>\n Java 17 sees the deprecation for removal of the Applets API via the JEP-398<\/a>, this one hasn’t been in use for many years, so it hasn’t caused much of a stir.<\/p>\n Java 17 also sees the deprecation of the Security Manager for removal via the JEP-411<\/a>. There was a lot of debate around this announcement, with admittedly a bit of drama<\/i>, such as when Apache Netbeans announced that it couldn’t support Java 17 when all it needed to do was change a few lines of code … JEP 406: Pattern Matching for switch (Preview)<\/h2>\n
\nstatic String formatter(Object o) {\n String formatted = \"unknown\";\n if (o instanceof Integer i) {\n formatted = String.format(\"int %d\", i);\n } else if (o instanceof Long l) {\n formatted = String.format(\"long %d\", l);\n } else if (o instanceof Double d) {\n formatted = String.format(\"double %f\", d);\n } else if (o instanceof String s) {\n formatted = String.format(\"String %s\", s);\n }\n return formatted;\n}\n<\/pre>\n\nstatic String formatterPatternSwitch(Object o) {\n return switch (o) {\n case Integer i -> String.format(\"int %d\", i);\n case Long l -> String.format(\"long %d\", l);\n case Double d -> String.format(\"double %f\", d);\n case String s -> String.format(\"String %s\", s);\n default -> o.toString();\n };\n}\n<\/pre>\nnull<\/code> case (in its two forms: statement or expression). Previously, a null switch variable resulted in a NullPointerException<\/code>. In its new form, we can add a null<\/code> case to handle nulls within the switch. Without case null<\/code>, the old behavior is kept, and a NullPointerException<\/code> will be thrown.<\/p>\n\nstatic String formatterPatternSwitch(Object o) {\n return switch (o) {\n case null -> \"null\";\n case Integer i -> String.format(\"int %d\", i);\n case Long l -> String.format(\"long %d\", l);\n case Double d -> String.format(\"double %f\", d);\n case String s -> String.format(\"String %s\", s);\n default -> o.toString();\n };\n}\n<\/pre>\ncase<\/code>. The following example from the JEP shows its use. By adding a guard to the Triangle case: case Triangle t && (t.calculateArea() > 100)<\/code>, we can create two cases: one for large triangles, and another for small ones.<\/p>\n\nstatic void testTriangle(Shape s) {\n switch (s) {\n case Triangle t && (t.calculateArea() > 100) -> System.out.println(\"Large triangle\");\n case Triangle t -> System.out.println(\"Small triangle\");\n default -> System.out.println(\"Non-triangle\");\n }\n}\n<\/pre>\nJEP 356: Enhanced Pseudo-Random Number Generators<\/h2>\n
RandomGenerator<\/code>, and a factory RandomGeneratorFactory<\/code>, which provide access to a random number generator implementation. The existing generators: Random, SecureRandom, SplittableRandom and ThreadLocalRandom; now implement this interface which adds, among other things, access to a random number stream (RandomGenerator::doubles(), RandomGenerator::ints(), …).<\/p>\n\n\/\/ the old Random generator\nRandomGenerator rng1 = RandomGeneratorFactory.of(\"Random\").create(42);\n\/\/ the default random generator, currently jdk.random.L32X64MixRandom but this can change\nRandomGenerator rng2 = RandomGeneratorFactory.getDefault().create(42);\n\/\/ shortcut for the default\nRandomGenerator rng3 = RandomGenerator.getDefault(); \n\/\/ stream all available generators and display their names\nRandomGeneratorFactory.all().forEach(generator -> System.out.println(generator.name());\n<\/pre>\n
Features that go from preview to standard<\/h2>\n
Features that stay in preview<\/h2>\n
A new port of the JVM<\/h2>\n
HexFormat<\/h2>\n
java.util.HexFormat<\/code> allows conversion of primitive type, byte array, or char array to hex string and vice versa.<\/p>\n\nHexFormat.of().toHexDigits(127); \/\/ \"7f\"\nHexFormat.of().fromHexDigits(\"7f\"); \/\/ 127\n<\/pre>\n
InstantSource<\/h2>\n
System.currentTimeMillis()<\/code>, LocalDateTime.now()<\/code>, and other date initialization with the current date.<\/p>\nInstantSource<\/code>, with a single implementation Clock<\/code>. The purpose of the InstantSource<\/code> interface is to be an Instant<\/code> factory. Instead of creating an Instant<\/code> with the current date, you create it from the InstantSource<\/code>. A test can then use an InstantSource<\/code> at a fixed date instead of today’s date.<\/p>\n\n public class MyBean {\n private InstantSource source; \/\/ dependency inject\n ...\n public void process(Instant endInstant) {\n if (source.instant().isAfter(endInstant) {\n ...\n }\n }\n }\n<\/pre>\nInstanSource<\/code> is initialized via InstantSource.system()<\/code>, and in testing via InstantSource.fixed(LocaDateTime.of(\/\/the hardcoded date time))<\/code>.<\/p>\nMiscellaneous<\/h2>\n
Map.Entry.copyOf(Map.Entry)<\/code>: allows to create a copy of a Map entry that is not connected to the existing Map.<\/li>\n\nProcess.inputReader(), Process.outputWritter(), Process.errorReader()<\/code>: allows access to the input, standard output, and error output of a process via a Reader or a Writer.<\/li>\n<\/ul>\nDeprecation and encapsulation<\/h2>\n
\nThe removal of the Security Manager was announced, because it is complex to maintain, expensive in terms of performance, and does not provide the necessary security in the face of current challenges. It was created to secure applets which, by definition, execute untrusted<\/i> code, and therefore no longer makes sense in a JVM that would no longer contain the Applet API.\nFor more info on its removal, you can read this InfoQ<\/a> article or this reddit thread<\/a> (I really enjoyed pron98’s responses).<\/p>\n