Jakarta EE JakartaOne Livestream Cloud Native for Java 2020, 15 May 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Fly Like a Rocket with Helidon 2.0 | JakartaOne Livestream 2020

Description

Tomas Langer- Helidon Architect & Developer at Oracle, Dmitry Kornilov- Jakarta EE & Helidon Team Lead at Oracle

---
JakartaOne Livestream Cloud Native for Java (CN4J) is a one-day virtual conference for developers, engineers and technical business leaders with the focus of building enterprise Java on Kubernetes.

This virtual event is a mix of expert talks, demos, and thought-provoking sessions focused on enterprise applications implemented using open source vendor-neutral Jakarta EE and Eclipse MicroProfile specifications on Kubernetes.

A

Okay, so I hope you can hear us we're live now with Thomas and mitri who are going to talk to us about Halden.

A

Let me see, do I have people in this session.

A

Let me see once again, hello, hello,.

A

Let me quickly share the screen before the top of the hour, just kind of to remind people that.

A

We need them to be involved and we would love if, um if.

A

You can join, but I want to show that the slides, so you can kind of see how to join and be aware that you can join us and follow us on Twitter on Facebook on LinkedIn and now that we have our YouTube channel.

A

Please follow the YouTube channel as well would love to to to have the traction there and on that same YouTube channel we will have all of our sessions uploaded as well. So that's that's one way to be in touch with us. The other way is to really get involved with Jakarta EEE and then become Eclipse Foundation member during Jakarta working group and then start working on the specifications.

A

All you need to do is the ECA and if you were available to attend the world's presentation earlier today, you know that he shared everything on his screen and we will be more than happy to share that presentation with you as well. So yes, please get involved and and and help us out on the journey with Jakarta.

A

So now, I will actually stop sharing my screen and invite mitri to share his. So he can start with that presentation. For today,.

B

Yes, okay, thanks Tanya! So do you folks see my screen I hope so yeah.

A

We see your screen so.

B

There, at the top of the hour, let's get started so welcome to fly like a rocket with halidom session. The main goal of this short presentation is to give you some details about upcoming Hilley down to the zero release. It's a major release containing many new features.

B

The agenda is that I, give you a short introduction using slides and the rest of the presentation will be demoed given by Thomas.

B

Safe harbor by Oracle, you should not make critical business decisions based on my presentation a little bit about myself. My name is Misha Kornilov I'm, a lead on project lead here are some contact. Details feel free to contact me if you have any feedback.

B

So this is how I looked before self insulation, and this is from how I look after that.

B

If you don't know what Kelly done is it's not an application server like Deb logic, upon needed, CPR, etc? If the Java libraries for developing market services like sprint boot quarterback's now, actually there are two chili dogs killing on SE and Hill and MMP. Two different frameworks use one of them in your micro service. Not both.

B

We created two hello dolls because they don't believe that it's a good idea to mix two development experiences carried on f-e is our reactive non-blocking framework. It's designed with performance mind it's similar to what Netflix is using for based services, it's similar to what Apple is using for its services. It's super fast and very small. It provides more than development experience with lambdas fluent API builders and so on. There's no magic, no magic means we don't use annotations where it's not dependency injection, but instead you have a full control of your application.

B

Mp is I would say a modern version of Enterprise Java, it's club-friendly. It supports micro profile and some Jakarta components it provides familiar cherry or spring would like development experience with annotations and dependency injection with magic, so killed on 2.0. This is what this presentation is about: it's not released, yet it's a preview, the religious plans in a second quarter this year, but they already released free milestones and the short milestone was released.

B

Yesterday you can use it to try new teachers and give us some feedback important to mention that he lead on two increases: the minimum Java version requirements. Now it's channel 11 inhibit on one. It was Java 8 in this version, Java 8 support is dropped.

B

Hit on is not only Oracle product in cloud they give shadow family aware are currently four products. Tip logic is our Jakarta application server offerings is our in-memory cache solution by the way the open source the origin of conference is going to be released soon to in the Dom. This is what I'm talking about now and Verrazano kubernetes based hybrid application management solution talking about kubernetes. They were two talks with tubular natives before mine, Sebastian and Adam used cabinets in their demos.

B

So it's an industry trend now and many customers I consider emigrate in their solutions to public or private clouds and embracing kubernetes as the development platform Oracle cloud native Java solutions are based on Cuban eight Eskew, so it allows customers to migrate to micro services architecture using their own pace and deploy it on premise or an oracle already out of public clouds.

B

Oracle provides enterprise, support for cloud native products and Healy done is not an exception. Hilla, don't support is included in support contracts for products listed on this slide, it's variations of their logical Cochrane's. So now, let's finally talk about halide on itself and new features and I'll start with Hill it on S II from this slide, I put all he'll Adonis seat to the JIRA features. Newly added features are marked green.

B

We already had a non-blocking web server configuration security and observability components, but these are missing some functionality to be called switch, complete, we're closing this gap by it. Reducing DD client, which is non blocking database access api, their client, which is non blocking HTTP api to call other services rate, your streams, reactive messaging. We have socket API, which was requested by many users and course support.

B

Now we have a full set of non-working api's, which you may not require to make fast and blocking service we're also introducing a new jail in base profile and kiloton common line, interface and I'll talk about these features later now, let's jump to heal it on MP the similar slide.

B

We support micro profile 3.2 fully, and this components are marked light blue. We support some Jakarta components. This can this boxes are highlighted, dark room. We continue supporting micro profile specification and adding implementations for a reactive streams in the active messaging.

B

We also aid in web socket to calculate socket, and course support and and B, but the main feature in Hillingdon and E's growl DM that if he would support Thomas real demonstrated in his demo, but I talked about the build profiles a little bit more, so we support now free form, comparison and he'll it dong-won we supported to in se, and only one in MP Sabattus. One is if you cutable jar file is fully supported from day one and it's a hollow jar. It means that it contains only application business logic.

B

All third-party dependencies are collected in a separate directory. It's done for better doctor layering support, and this is what Adam Dean was explained in his talk. I'm, not going to repeat it. Using this approach, you can build your docker image, it's really fast and upload download stuff faster.

B

The second profile is J link. It's new in Hillard on to the Giro you know, J link to it was introduced in Java 9. It's used to build Java Runtime environment, which contains only functionality requested by your application. It makes your overall image size smaller. It also optimizes your application startup time and what's very important in, doesn't introduce any restrictions, their application, so you can call using em everything what Java provides and the last profile is drawing native image.

B

We support it in mp, since version 2 the zero, and it allows you to build an executable native image, gravure native image, both based on your application, with advantages of almost immediate startup time and the lowest possible memory consumption as a drop back. It introduces some restrictions to application related to usage of runtime Java features such as reflection proxies and so on. It becomes important than instant third party libraries. The channel grow them optimized.

B

Here is a slide about better understanding how these images are related, so I created free images based on our Hill Adan and P QuickStart application and sorry chunk too early. So here is a docker image size comparison. So you see that char profile image is the biggest is to con, given something megabytes it because it contains the full Java, runtime environments. J link image is smaller. It's just under 200 max it's optimized and growl, the M natively, which is the smallest and startup time.

B

You see the same picture so the startup time for which our profile is the highest. It's almost 1.6 seconds. It was another fast computer. By the way, J link profile is one second, something and foreground way and making, which is almost immediate, but on the other hand, what is 1.6 seconds startup side all right. It is almost nothing comparing to the time of port provisioning and to turn edges which may be like 30 seconds or something like that.

B

Okay, now jump into the second feature: elee, don't CLI it's a common line of TDT created to make developers life easier and more fun. It simplifies project creation. We have several Sen and beach in place choose from.

B

It also has a development port which tracks your source called recompile seeds and reruns the application, if it detects some changes without like restarting 20 convenient they in the highly requested by the community. It's doing active development. We are planning to add more interesting features soon. I don't want to spend much time here. Thomas wrote demonstrated in his dama to DB client.

B

He laid on the decline is non-blocking database taxes API, for he laid on s e.

B

When we are developing a non-blocking service accessing the database was always a challenge, it's mainly because we are not too many non-blocking database drivers and agent blocking database access functionality to your non blocking service doesn't make any sense. So if the decline is not change anymore, it adds non-blocking database texas api and allows you to use existing blocking database drivers in a non blocking way sounds little bit complicated again. Thomas will demonstrate it if he casts time in this demo. So digitize API supports relational and no sequel databases.

B

It supports all healthy Donna mobility features such as metrics health checks tracing. It doesn't introduce any new data based access to language that the base native direct is always used.

B

It uses convention over configuration in a way that all database statements using your application, I externalized in configuration files, not in the code, which adds limited portability, so you can switch databases by changing, only configuration without three comparing your recompiling, your application, their client, another new API Hill it on 2.0, it's a non-blocking, HTTP client, it's designed and tailored on assyria active style supports all HTTP operations provides full integration with observability features and as follows: redirects also can be extended by extra functionality. Thomas will demonstrate it in his DMR, get reactive streams and messaging in.

B

In the previous milestone release in m2, we introduced three active stream, separators and reactive messaging support in both ACN n P in NP, we're extending our micro prophet Exley's by agent implementations of the corresponding microprocessor specs in SES done by 18 new api Sandhill ontology style.

B

It has a actually huge sense for a see because reactive streams processing is must-have functionality for modern non-blocking frameworks.

B

By the way, reactive streams, implementation is contributed, charts by David Carr knock, who is a famous guy in reactive job of work. He is a father of project director, free.

B

We have a goal of delivering one of the fastest reactive steam spray operators. Implementations and here is a comparison for implementation with small right small rise, a reference implementation of microprocessor. You can see that we are about twice faster, which is perfect result for us, and the last feature I'm going to highlight course.

B

What is course, you know browsers by default by using same origin policy, so browser can access resources only from the same region that provided the app means that you have your application on the solar. You access it by IBM browser you can't access any other resource other than located on your server, which sometimes is a little bit of a problem and to solve this problem exists, cause originally search sharing.

B

This is the stuff which controls access by their up from one region to resources from another. It uses HTTP headers for that and in Hillingdon we support it fully in SE and MP, and it's automatically supported by Hillingdon features like health check, metrics and open api. So for better understanding. So this is a slightly explaining how it works so suppose.

B

Hospitality dot-com hosts an application but retrieves friendly messages from other apps and we have a hill it on QuickStart application, write it on my server calm, which provides some greeting right. So in your browser browser load, the del come up from hospitality, calm and hospitality. Calm becomes course origin.

B

This application request something from my sovereign calm. If does get on great results, so host becomes my server calm and browser course. Implementation adds host in the region headers to your request, so solar receives two requests purposes of a little application and Helen own application checks. Is this restore and these methods allowed for resource sharing? If so, that entity returned, if not and for free? Is it job? So it's as simple as that, how its implemented in F II.

B

You create a course support, basically build it using a builder, and you add, across the region, support to the Builder. The first one allows to trick that and allow the region's the hospitals that comment. Other calm and cross written support, dot create basically allows everything else to other origins.

B

After that, then this course support is creative. You just register it the same day as in your routing, important that it should be added before arranged in the great service which needs it works for in empty. It's even more easy. It's just annotation across the region. There you provide all the required information.

B

Okay, nineteen minutes I supposed to have enough time for the demo. Thank you very much and I'm passing the control to Thomas.

A

So before you do that, maybe just a quick question: is there a list of JDBC drivers that are native image friendly.

B

No, that's actually a good question, so where are some drivers which are native image friendly? So we faceted on h2 right and there are some other drivers provided by Red Hat, which are native image friendly and what I can say that Oracle works on make it or it will be be driver native image friendly, but unfortunately I don't think that it comes being scale it on to the general release. It may come later on this year. I don't know exactly today.

A

Thanks for the question, Ellis and, of course, metric or answer, and let's see Thomas if you can share your screen, okay, it's good very.

C

Good, alright, so I will start by using our Haledon come online interface to show how the breaks, so, let's just show that I'd a little bit it's it's a single file. Download I will talk about it a bit later when I have a few moments during our built. So first thing that I want to do. I want to create a new holiday application. So what I'm doing here, I'm creating a version two muscles to the application with sesame flavor. Of course you can use MV flavor as well. Now this will create an application.

C

If I look here, you can see that it is the usual hell, didn't application prepared with the dockerfile llamo file for Copernicus and the sources now. The next thing I can do. I can build my project, but because I already have Holden running I just want to change the pork. So I will do a quick figures here. I will change the port to something that's not worked now. I will do how we can build and I will give it make a cloak. This is an example of a head of time: compilation using Oracle, growl VM.

C

We support it as Dimitri said, both in SC and then MP I'm, showing it on se, because it's faster and it doesn't block us that much now I can talk about the.

B

C

Tool itself, it's actually built using native image as well. So it's really quick to start and it's written fully in Java.

C

The good thing in Haledon is that all the things that you see here, building of the application building of the native image and building of a docker file, a docker image, are done using the usual tools that you would expect. So you can use pure maven. You can use the chorale vm native image to compile it if you provide all the switches and you can build docker using the docker command. So as you can see I my native image is built for Hilden SE. Now, I can just start it. You can see.

C

Put it really really fast running on port 8080 9 now I wanted to show a few things here that are interesting when it comes to size comparison, we talked about it a lot and the price of resources was discussed many times.

C

So we can see that the full size of the native image is 23 megabytes, and this is a an application that contains everything that you need. You don't need JB JVM, you don't need anything else. That's it. We can compare it to the actual size of the libraries that are used by how atom, so that would be this directory for se.

C

You can see that the whole sum of all the dependencies of Houghton I see is 5 megabytes when you use the basic configuration and of course we can have a look on the the size of the application itself, which is 12 kilobytes, it's the whole jar dimitri talked about it. That's why I wanted to show here I will show something some parts of the CLI later so, let's, let's finish this screen, and let's go to my head and application.

B

C

This show I have created using my Haledon init command and SE application, and an empty application for Culloden I see I am using what we call a dev mode. You just do hold in death. It will start your application and whenever you do a change, it will make sure that you get the latest version in Europe in here. So you don't need to restart it manually. It's all done behind the scenes, for you.

C

I also have an empty application, and here I will just show what I did before we started, because this takes a bit more time. So I have built the native image for our MP application. I will scroll down to the important stuff. So here you can see that the build time is 2 minutes you.

C

There is no free lunch, you pay for stuff. So if you want to have a head of time compilation, you obviously need to build that a compiled file, and that takes time and the bigger your application is the more dependencies you have the longer it would take. Of course, you also need a bit of memory to be able to compile it. Now. The application can be started, I will just start my empty application and now I can go and show what's actually in this application. The start of time here it's printed here, it's 55 milliseconds.

C

There are a few things that must happen when you start an application in native image, such as loading of your configuration, because you don't want to be stuck with the configuration you had at the time you compile your application now. This is the source code of my Holden and B, and what's really important here, this one cost. This is our jax-rs resource, and this is the only thing that we need to have a working, jax-rs or rest application in Haledon NP. It uses the usual allocations of jax-rs.

C

You can also use annotations from city eyes, such as inject, to create a new instance of this resource, injecting our greeting provider and, of course, that uses micro profile, because we are an implementation of micro profile. So.

A

C

We are injecting config property when you build this application, it uses a main path that comes from Holden and it just boots the CGI which takes care of everything. So obviously we need to have our business dot XML file, and then we have the micro profile, config properties to set up Haledon. So here what I'm doing I'm preparing the server port the listener address and then I'm configuring, the service of the tracing, because I want to also show tracing later on. So I think that there was a lot of stuff said about micro profile.

C

So the last thing I will do with that is I will show how we handle the docker docker image and, of course, I am I need to be in the MP.

C

And I- don't have my history here so come on. This is this is what always happens when you switch around too much now here we are so I will build a docker image from my MP application.

C

Let's see if it it worked, you can see it's quite fast because of the layering that we talked about. I have changed nothing in my application, so it didn't need to recompile and it just created a docker image. The good thing I, think from from for users of docker, is that it's actually using the total command. So we don't have custom maven plug-in. We try to use the tools that are built for the task and for building docker images. We think the correct tool is toker.

C

Now, let's go back and start talking about the new stuff in Holden se, because that was the most complicated part, so I will switch to hold on I, see and what I prepared here is an application that uses a database and it calls another service using our reactive client. So I will start from configuration.

C

Supporting again, we have the tracing configuration to be able to see the two services that are part of our communication chain. I have to change the port, so we don't have a conflict, so our front-end is running on port 80, 60 and now comes the configuration of the database, and this is what Dimitri talked about. We try to keep the database configuration solely in your configuration file.

C

So if you decide to switch from H to database, which is an example here to Oracle database, for example- and you need to change your selects, you can just do it in the configuration file and the application will work the same way. So here you can see that we configure our collection of Culloden, actually supports encryption of keys in the in the configuration files, both in Helen Sdn and Helen and B.

C

So if you wanted to do that, you can, of course you can also use kubernetes secrets and load those files as an additional configuration source. I will here you consider statements. We have wondered this kind of system statement that is used by health checks. If you have them enabled and then I have just a single selling here, that is using- that's called a good greeting that will select just something from the tape from the table in our database.

C

Now, that's it. Let's have a look. How that looks in the source code, so I was close, our MV resources and let's have a look at our service that this responsible for handling it. So I will start with creating our web client and TD client. So, if you have ever played around with halidom, you may have seen that we like builder patterns and fluent API, so we try to be to allow you to build everything in that in a flow. So here we create a builder.

C

We give it the configuration of the database from our configuration file I configure an interceptor which is a deep database, counselor so visible at metrics into our database and I. Just build a client and I will immediately show how we build the web client, so here I'm just using a base. Uri hard-coded the web client similarly has a possibility to configure this using our configuration. I just didn't want to do it, I want it to show one using programming and one using the configuration.

C

We add the support for JSON processing here and again, I'm registering a counter and then building my client. So this is the first step we prepare it when we prepare the instance- and these are reusable so later on- we use them in route. So let's have a look, how the routing works. So for those who haven't used thousand as E, we have the concept of routing where you configure rules on a specific route. So, for example, this service is registered on slice Creed and it registered its own subpaths for various HTTP methods and policies.

C

So here we can see that if you call get requests on the root of this service, we have some default message handler. If you call get requests with a name parameter in the past, it can call the get message handler. That's the one I will talk about, and then we have a put support on slide greeting that will update the message, so the one that I played around with is the good message handler, which normally, if you create our QuickStart, will just print hello and the name that we have sent.

C

What I changed here is that I have decided to call the database first and see if we have something stored in the database for that user. So here you can see that I try to execute a database statement, and here we have the name to get statement which uses the category king name. So this is the mapping to the configuration file and I gave it the parameter.

C

It's the name that I have received. So if I go back to the configuration, we can see that I'm using the usual JDBC approach of a question mark instead of a parameter. We also support named parameters. So you could actually do something like this, but then you need to provide a name map of parameters when you may execute your statement, so this name get is returns a completion stage because it's a non booking environment.

C

By the way, it's important to understand that for JDBC, which is working API, we obviously need to do something about it, so this is handled by using an executor service. So, even though it's non booking, there is still a set of threads that will book on the JDBC calls, but they will not be booking the server and they will not be booking other routes that are executed in this server.

C

So when I executed, I get an optional with the result, because I get is basically a statement that returns either thing or a single record, and if it's not found I just returned a 4/4. If it's found, I will invoke the other service and of course we need to handle exceptions. So if you, for example, if it returns more than one record, something has to happen so I just send the exception over the wire which either sends the stack trace over the network or just a 500 error, depending on configuration of web server.

C

So here what I am doing, I'm just reading the call number 2. You can also read columns by name from the statement, if you generally don't like ordered parameters. But it's the like. The simplest approach is just to use the ordering number and I read it as a string and now my call MP of a method is the one that is invoking the web client and calling our Alden MP service and here for the web client I'm just saying that I want to invoke a get request.

C

So does the HTTP work I'm, giving the past, which is the name that I have retrieved from the database and I execute the request? And again this returns a completion stage, because the request is happening, a synchronously with a non-blocking approach when it comes back, I just copy the content from the outbound requests, so I copy the status, then I copy the content and that's it and again exception handling. So this is a full reactive flow of curing the database and invoking web service with the result of the database.

C

Now you can also see one important thing here. Reactive programming is in general, a lot of scaffolding and not that much business logic. If you compare it to synchronous programming, so you get a huge amount of performance out of it parallelism, but you also pay for it. With the complexity of writing such code and I'm really looking forward to project loon, finally hitting the JVM, because then we should be able to have synchronous in occasions with without the price of walking threats all right, so the whole thing is set up.

C

I have my hood and you're running in my head in death mode I have am be running from native image, so we should be able to actually invoke it. So, let's do, as do our color command. I have single record in the database, which is for Jack, it shows returned Sparrow, and so it does so now. You can believe me that we have clearly the database and called the remote service. I can prove it. We have tracing.

C

There was a lot of stuff set about all the observability features and tracing this one of the most useful ones. So what I can do now? I will just search for old traces that are in my system, and you can see that there was an S II call. So there was an HTTP request coming to our holida nestle on slash, agree, slash check, which is what I have done using Carol command and now I can show that whenever you use a Heligan feature, we always think about metrics and about tracing.

C

So when there is a database, you have the possibility to hook it into the tracing chain, and you can see it here. So we have seen that there was a get operation in the database. There is the select that was executed, of course, without parameters we have the database type and, if I show all our records, it also says that it returns returned a single result.

C

Now the next step is our HTTP client. So here this was the get request called to a remote server, and here you can see that we have switched contexts. We are now in our MP server that is processing this query and handling it now.

C

As usual, the first request is slightly longer than than the subsequent ones, so I will just show how it looks when you have four top applications, so it's 12 milliseconds, look much better for another bound call.

C

So that's the tracing art and the next part is metrics. So let's have a look on on how metrics web so I will just do a curl with photo metrics and we.

B

Can see that we have two.

C

Metrics registered there is the counter for localhost, it's form coming from the web client, and we have the DB counter forged greeting, where I have decided to have for each statement that we could have. We would have a different country, so here we can see that we called the gate on localhost was a response 203 times and we called gate greeting three times as well. So this was all my oh.

C

My commands I could of course, try.

C

Existing one, so we can see we got the 404 back and now, if I look on the best-managed to type correctly, we can see that the numbers are not different because, of course the get we think returned no results, so we didn't even call the remote service.

C

So this was the how it binds into our metrics. So you can see that with both metrics and and tracing is working. One thing, I didn't do I didn't actually at health check support for the database. We have it I missed that one. So if I do our command, but this time I will request health.

C

And you may have seen that all these commands are repeated over and over again during today, and that's because obviously this is all coming from micro profile specifications. So even though we are using Holden s EE and you can see that we have a few checks that are registered by hand I have actually, these are like built-in health checks in Haledon that you can use if you, so we have deadlock, disk space and memory health check. So even though in Helen s II, we we provide a reactive implementation of libraries to build your micro services.

C

We have adopted the Micro profile, KPIs that are not CDI specific and we are using them as well. So each basically each hidden feature that you can see here is somehow connected with the micro profile API. There is one exception which is metric, which is config because configuration we have designed our configuration before micro profile came into being, but for metrics the format that you see you can also do. You can get the JSON format of Micro profile or the text format of committees. So I can show that here if I go to.

C

So it's 80, 60 and I will do metrics just application. You can see. This is the usual promises for month that you get by default.

C

Or you can request a JSON format, which is micro profile. The same thing is for health checks. We have adopted the health check format of micro profile, even in health and se, so it doesn't matter whether you choose to use evident MP or held in SE. The observability features are the same, so you can use the same tools to monitor your application to troubleshoot your applications.

C

Another thing for tracing actually tracing is also part of micro profile specification, although it's basically take using the open tracing specification for all the api's, except for one on location and if you use hidden se or MP, they both propagate automatically or the required headers for for tracing. So if you have, you know incoming headers from zip in any music in yourself, and then you call another service that uses a pin. You will automatically get all of this in your second tracer. The same comes from for Jager as well. Obviously,.

C

All right, I think that I have guns through through all those things. I wanted to show you, as you have seen, the native image worked the same way as if I launch the application using java, so the micro profile in native works.

C

We have tried to limit the amount of things you have to adhere to to be able to use the native image, but obviously- and that was already set as well native image is tricky, because you need to be sure that everything that you do will work in a native environment and ahead of time, compilation using growl record, for example, thus all static initialization at the build time.

C

So, even though you can use almost everything from java, except for method handles and a few other things, you need to be aware that whatever you do in a static, initializer will be initialized at the build time. And then, when you run your application is just there and you can do nothing about it. So that's the reason why we differentiate as the square goes obviously between a builds time in the run time, and you need to be very much aware of what you do in which of these areas.

C

So if you build your own CDI extension, for example, you need to make sure that you initialize things at the right time and with that I have gone through the demos. So if there are any questions, I think we can, we can go to that or.

B

If there is anything.

C

Else, I, you want to see from all of them.

A

Please now it's your portunity for questions and mitri. Thank you for answering Mary's question very good to see that you're here. So ok.

C

I can see it now. So let me answer sorry about it. I missed the question, so is how Eden underlined reactive stream support.

B

C

Or reactive messaging, so I think the materials are there. We have our own implementation. That is neither so that's that's how we we decided to do it in our case for the JDBC drivers that were mentioned before. As far as I know, there is no driver that works without modifications we are actually using. In our example, the Quercus HT 2 driver and thanks for guys from Farkas to actually doing that work because it's it's demanding and you know with every new version you need to go back to it.

C

The only driver to time managed to use I think was the reactive driver with DB client that worked without additional changes in in native image.

C

Ok, thank you very much for attention then, and enjoy the rest of this conference.

A

Yeah, so we are going to start well, first of all, Thank You, Dimitri and Tamas. This was great thanks for participating and and supporting and being part of the Jakarta on livestream for others. We are going to start the next session at the top of the hour. So I can see you there, but I also want to take the opportunity and just to share my screen briefly and if you're interested give you a little bit more details on how we can how you can further help us out so and what other news are out there.

A

So I told you to get involved. I already mentioned earlier and I. Think well was mentioning earlier that we're preparing annual Jakarta 1 livestream in September so keep your eyes and ears open for that for Portuguese speaking people, please be aware that August 29th there is going to be another Jakarta one live stream. If you are interested in organizing Jakarta, one live stream in your local language.

A

Do not hesitate to get in touch with me, I'll be more than happy to help you out, and then we are organizing eclipse con 2020 as a virtual event, and we already know the dates so please register for the event and and and mark your calendar for October 19 to 20 seconds. You already know how many compatible implementations we have out there. So you can also visit the on jakarta ee website to the compatible implementations and read about them. Let me see what else can I tell you?

A

I am hoping that you're all familiar with adopt a spec program for drugs. So if you are a member of any drug that is particularly interested in one of the specifications, please let us know we would love to get in touch and officially put your logo on the Jakarta website.

A

Subscribe to our YouTube channel.

A

Projects and community mailing lists and I will stop with you know, highlights or focus from the on the Jakarta e9 release. We already talked about this earlier or well. Brother talked about this earlier in his talk, but you know wanted to to share again and invite you to get involved. We still need a lot of help with the change of the Jakarta or Java clean space rather to Jakarta.

A

There is still tons of work to be done and we are looking for more participation from the community.

A

You see. Are there any questions.

A

No, no questions for me. I didn't necessarily.

A

One thing that I could show you is actually.

A

So one more thing that we just put out there for people who want to get started with Jakarta Yi: this is being developed, but for the time being, you can also see some of the examples and starting point projects. So if you're again interested please get involved there and you let us know if you have any feedback and help us further develop. The true trial started for Jakarta, II and I, guess that is it any more questions comments?

A

If not everyone? Thank you. So much for attending and we'll see you in the next session, thanks Thomas, Tigers, meetry, again bye for now, Mike.