Java 21 API New Features

Learn about the many enhancements that Java 21 is giving you in the classes you are using every day.
Java 21 is introducing several major new features: virtual threads and pattern matching. But it also brings new many small additions to classes that you are using every day, and that you can leverage from day one. This video covers the most important ones: the String class, regular expressions, the Collections framework, some additions in Java I/O, the Math class, and, as a bonus some important deprecations you need to be aware of.
⎯⎯⎯⎯⎯⎯ Chapters ⎯⎯⎯⎯⎯⎯
0:00 Intro
1:18 New methods on String, Character, StringBuilder and StringBuffer
2:34 Naming your capturing groups in regular expressions
4:01 Additions to the Collections framework
6:22 Localization in the Date and Time API
6:53 Additions to the Java I/O API
7:49 Using AutoCloseable with HttpClient, ExecutorService, and ForkJoinPool
9:29 Additions to the Math class
11:25 Multiplying BigIntegers in parallel
12:00 Additions to the Thread class
13:25 Additions to the Future interface
14:45 Deprecation of finalize()
15:42 Deprecation of the constructors of the wrapper classes
16:20 Outro
⎯⎯⎯⎯⎯⎯ Resources ⎯⎯⎯⎯⎯⎯
◦ Full #RoadTo21 Playlist ➱ • Road to 21
◦ Java 21 New Feature: Virtual Threads ➱ • Java 21 new feature: V...
◦ Java 21 New Feature: Sequenced Collections - JEP Cafe #19 ➱ • Java 21 New Feature: S...
◦ The JEP 444: Virtual Threads ➱ openjdk.org/jeps/444
◦ The JEP 431: Sequenced Collections ➱ openjdk.org/jeps/431
Tags: #Java #Java21 #Tools #OpenJDK

[@java]

We got 2:34 wrong. Named groups in regular expressions are supported since Java 7. What's new (in 20) is that the methods related to named groups moved up the dependency tree from `Matcher` to the interface `MatchResult`.

[@zoladkow]

no worries, everyone here knows the vids are just about coffee sipping with some words inbetween 😁

[@brandpcalderon5343]

i was going to complain the same, named groups are allowed since I learnt regex like 3 years ago and more I suppose

[@OzoneGrif]

Yeah I was really confused about this. Was going to comment the same thing. Brain fart !

[@onebeartoe]

Thanks! I cant wait for the named String groupings to help parse flat files.

[@boyneverstop]

Thank you for making this video! Very useful

[@JosePaumard]

Thank you!

[@OriginGaming101]

Came here for Java, stayed for their use of my favourite local cafe in the background! 😂

[@Talaria.School]

Great topics I had to comment and thumbs up.
Thanks a lot José.

[@JosePaumard]

Thank you Khaled!

[@svalyavasvalyava9867]

Awesome video, thank you ☺️

[@codeZarathustra]

Good explanation Sr! Thank very much!

[@JosePaumard]

Thank you!

[@YvesDautremay]

Thanks for the overview ! A few nice things that may be shadowed by virtual threads

[@JosePaumard]

Thank you!

[@ajaythombare6235]

thank you for valuable information sir...
btw I like ur coffee cup

[@nO_d3N1AL]

Really appreciate this format of doing updates. Also a cool idea: the amount of coffee left in the cup should match the progress of the video ;)
Also thankful that new APIs and methods aren't using checked exceptions! 🎉

[@JosePaumard]

Well it does ;) This is real coffee that I drink during the recording.

[@nO_d3N1AL]

@@JosePaumard Cool! I bet it's actual Java coffee too ;)

[@JosePaumard]

@@nO_d3N1AL Oh yes it is! 😊

[@computer_science10]

thank you

[@radoandrey2268]

Thank you

[@ahmadfx8962]

رائع

[@rishiraj2548]

Thanks

[@si2001si]

Hi Jose, Big Fan!! Your Coffee looks yum

[@1gmar]

Apologies, isn't the feature mentioned at 2:34 already available since Java 7?

[@mikolajpodbielski]

Yes, it is.

[@christofferhammarstrom]

What is new in Java 20 though, is that MatchResult now supports named groups. Perhaps there was a mixup here.

[@JosePaumard]

You're absolutely right. What has been done in 20 is that the methods are now available on MatchResult (interface implemented by the Matcher class), along with some others. I should have mentionned namedGroup() also.

[@RishiRajxtrim]

🎉

[@hephestos300]

after what JDK will be the whole lambda as was introduced in JDK8 deprecated and ousted ? Can't wait the moment

[@edmaphis9805]

Emoji methods !!
Holy smokes !!

[@jeromemainaud]

3:59 With large patterns in a multiline string, instead of using backslashes you can also use the COMMENT flag. This will be more readable.

[@TheNeoxpert]

08:44 close() method will be called no matter what? what if there is System.exit(0); in the try block?

[@JosePaumard]

Nope, not in that case. If what you need is to kill your JVM as fast as possible, taking the time to execute all your close(), finalize(), etc... would probably be a mistake,

[@mortensen117]

So, named groups, which arrived in Java SE 7, is now suddenly something new?
If you could write the name of the group in some meaningfull way without of a lot a string gymnastics, then there would be some thing new.

[@JosePaumard]

What is new is that the methods are now also on the interface.

[@ClydeXT]

In your example at 7:15, your types should be "byte[]" instead if "int[]"

[@zoladkow]

ahh... certainly too much coffee 🙃

[@oifbert]

What's new about pattern matching? Named groups have existed for years.

[@danthe1st]

Shouldn't it be "Charset" instead of "CharSet" in 7:09?

[@muzzletov]

CharSet is just short for Character Set. So, no, it shouldnt.But yes, the class is called Charset, which feels wrong :D

[@danthe1st]

@@muzzletov If the JDK calls it Charset, it is Charset and not CharSet. And honestly, I would consider charset to be its own (atomic) word and (at least in the context of Java,) it isn't really a set of characters.

[@JosePaumard]

It should. Thank you for pointing it out!

[@Speiger]

IMO the state enum is missing something.
"CREATED" or "UNSTARTED", because there is no way to track if a task was already started or not.
Which is useful if you have a queue of tasks that are expensive and you want to cut basically anything that isn't started yet, and you didn't expose the Queue itself.
Unless state returns null if it isn't started yet, then i take that back.

[@pravinjain2834]

I fully agree, currently it seems they put the state as RUNNING as soon as the task is submitted.
probably it is being thought that since we can have virtual threads, the number of threads in the pool will be more and no task will be waiting to get started. but for tasks which are computative (no blocking activity), one may not like to use virtual threads.
In any case you are right, that state has been missed.

[@edmaphis9805]

So Autoclosable is a functional interface?

[@JosePaumard]

It is. I'm not sure that it will be very useful to implement it with a lambda though...

[@edmaphis9805]

Ok.
class Closer {
static void closing(Closeable closeable) throws Exception {
System.out.println("Now closing: " + closeable.toString());
closeable.close();
}
}
class AClosable implements AutoCloseable {
@Override
public void close() throws Exception {
System.out.println("I am closing");
}
}
public class CloseableDemo {
public static void main(String[] args) throws Exception {
var aclosable = new AClosable();
Closer.closing(() -> {
try {
aclosable.close();
} catch (Exception ex) { }
});
}
}

[@edmaphis9805]

@@JosePaumard Thanks, I tried anyhow. 😉

[@JosePaumard]

@@edmaphis9805I'm not sure I would use this pattern in a real application 😉

[@akaalkripal5724]

How's the coffee?

[@JoshuaMuzaaya]

wow, all these have been in Erlang/OTP, Python, Elixir right at those languages' first version.

[@zenmony-dot-com7457]

Pleasant coffee lounge atmosphere, perfect for learning.

[@JorgetePanete]

I think close() should be called automatically when the object is no longer needed

[@SourabhBhat]

Wouldn't that cause the same problems as finalize?

[@JorgetePanete]

@@SourabhBhat At least if it's called at the end of the scope of the object I think there wouldn't be problems

[@alessandroautiero5414]

​​@@JorgetePanetein a method's scope that's probably easy enough (GoLang has the defer keyword for this use case pretty much), but how would you do it in the case of a field? It would have the same problems as the finalize method I'm pretty sure.

[@JorgetePanete]

@@alessandroautiero5414 I don't know, I just wish it was like Rust does

[@mattidragon835]

@@JorgetePanete the reason rust can automatically close is it's strict borrowing and moving rules. In java you can move objects freely and rely on the GC to clean them up

[@sneibarg]

Naming capture groups is a super old feature of Python.

[@Parkhill57]

piano in background still drives me crazy.

[@kitkarson4226]

Why do they pollute Character class with emoji methods 🤮? They could have made it as a separate util class .

[@pravinjain2834]

Yes, there could be a separate class for all the Unicode properties. There are still a lot of unicode properties not available in Java API (Character class).

[@nicholas1460]

You used var. Shame on you.

[@JosePaumard]

Oh no, I love var.

[@nicholas1460]

@@JosePaumard var var var var. Just for you. Quess what objects I'm using.

[@JosePaumard]

@@nicholas1460 You don't need to use it everywhere. But there are still many places where it will make your code more readable.

[@kitkarson4226]

var is a great addition to java. I have been using for couple years now,

[@khmarbaise]

var often shows that the variables names used are not good enough to transport the information. That lets me reconsider my names and make it better (hopefully ;-))

[@mortensen117]

AutoCloseable?
Why are you presenting these OLD features as if they were new? Are you crazy?

[@kitkarson4226]

Watch it one more time. AutoCloseable is OLD. But now ExecutorService, HttpClient..etc also implement the interface. Thats what he says!

[@JosePaumard]

It's not because it's old that it's not useful. With the deprecation of finalize(), AutoCloseable becomes more important. Thus the new classes that implement it.
Keeping repeating old stuff is not a problem for me.

[@Kjhd9987hy]

What an old ugly mess language. I feel bad for android devs when they gota look at swift and swiftui