 All right, so I'm Mike DiPaolo. My day job is I'm a system administrator. In my free time, I began shooting to X2Go and numerous related and upstream open source projects like Live SSH and Pulse Audio, et cetera. I've recently been working on better and greater in X2Go, Tiger VNC, and other remote desktop solutions into the footwear desktop experience. But before I begin, I'll talk about this presentation. It's not about VDI, it's virtual desktop infrastructure. I mean, the assumption of VDI is you have to have a hypervisor, and you have all the special code to make the desktops possible. Well, it offers features that are really nice to be like, here's this desktop image that users get, here's the system image that users get, if we've inch you from remoting into your own desktop or laptop in the office, or from going to a physical server in your lab, et cetera, or into cloud instances where you can't do nested virtualization. If people have questions about these, I can talk about them, but they're not part of this presentation. What this presentation is about is remote desktop. So I mean, most people think, when they see remote desktop, they think Microsoft remote desktop protocol specifically, and their client is called RDC remote desktop connection, whereas as people on PR talk to you, you believe the best generic term is remote desktop. Some people call it like terminal server, but that can be for other things too. It's also really like antiquated term that you have to explain to less-singled people. So that's why I think the best term for multiple, well, the protocol of the solution is remote desktop, and that generally includes remote applications too. And the three solutions that I'm, the two solutions that I consider stable and ready to be used right now in Fedora are Tiger VNC next to Go. Note that Tiger, there's numerous VNC servers and clients and mutations out there, but Tiger VNC is the best, and most well-rounded VNC server and client for Linux and the clients for Mac OS X and Winners are pretty good also. So Tiger VNC is actually included in well, but Fedora has a newer version available with features. And X to Go is in the EPL in Fedora. Spice, I'm not supposed to EPL by notion of Fedora. And again, so X11 itself, X11, X to Go is a display server. Display server handles input, keyboard mouse, newer devices, graphics, display, and some other stuff like footboard sharing and window manager properties. I mean, there's really good presentations, by the way, and they develop really about the limitations of X11, but X11 itself does not concern things like audio or what applications get stored up automatically when they log into desktop environment. Those are all separate standards that are related to X11, but not port of X11 itself. So I'll put these quick-store guides available somewhere like on the Wiki, but it's really simple to get to X to Go server started. It's just install those two packages, install compatible desktop environment. I'll explain more about this later and then enable those two services. And Tiger VNC, it's one server package to install. You can basically use a needed desktop environment, although the logic for launching which desktop session is currently in that script right there on the Fedora systems. And while you can often get rid of the simpler command line, command I recommend, this is the one that I've always been using successfully. So it's a VNC server on the next available ports and you can connect to it with your own username and password because it's using PAM authentication. And by the way, you will need to open up the firewall port. So if it says I'm running on display colon two, that's X11 port colon two and VNC port colon two. So it'll be 5902 for VNC to open up your firewall. So that's the intro and I'll talk about the depth of the different features of the remote desktop solution. So like I mentioned, Tiger VNC, it's in RAL, it's in Fedora, and X2Go is, you know, it's relatively new to Fedora, which only about two years, included through about a year and a half ago by Orion who's been an excellent packager. And so the main difference in spice is also available too. I mean, the main purpose of spice is that it's a protocol that's the KVM hypervisor influence as a better alternative to VNC, it has a lot more features. But there is that you can run spice for like, remote desktop, there's Xspice, which is basically, you know, X.org configured to use spice as like a driver. However, there's really like poor integration, like right now there's a bug where you have to like tell it to use, use or live exact, like Xorg.vl instead of use or live exact Xorg or something. I intend to fix those bugs soon, but for now I'm only recommending people try Tiger VNC and X2Go. So you can see right here, so the very difference between Tiger VNC and X2Go is, you know, Tiger VNC does support the bad they needed to feature, I mentioned earlier that it's really this resume sessions so that, you know, if you lose your connection to your remote desktop server, your apps don't just like stop running. You know, you can reconnect and resume your state with reliable Wi-Fi like this. The ability to resume a session is especially important. People talking to the Hyatt Hotel here is unreliable Wi-Fi. So, I mean, Tiger VNC itself, it's basically, you know, it's a little bit more than just remote display server. There has been significant, you know, development in terms of improving performance, especially in terms of, you know, better JPEG compression and adapting the algorithm JPEG compression to the type of data that's on the screen. But it does not support the other features that you associate with the Linux desktop like audio. It does not support file transfer from client to server or printer sharing. You know, there are other Tiger VNC clients. I'm sorry, there are other VNC clients out there that you can use with it. And those are available from mobile operating systems, but there's no official Tiger VNC client for mobile operating systems. And the thing that's actually going to, you know, it is like, it does have those desktop features that you would expect from some of them Microsoft remote desktop, for example. And I'll talk a little bit more about when performance later, but the really nice thing that actually is going to spice out that Tiger VNC lacks is going to the cache like image data. So say you view a photo or you open up your start menu once and then you close it and we open it. With Tiger VNC, it has to be sent the data all over again. Where it's actually going to spice inside the in-memory or on-desk and it can just show it. And the extremely, another perfect example of that caching is moving windows around. Simply dragging and dropping, dragging the window with the window manager. The Tiger VNC, the image data has to be sent where it's actually going to spice because it's like moving the window out, you know it's contents. And these are more like the first page, like the high-level features, these are more like the more technical features. So I said I mentioned the integrated session handling. So Tiger VNC, it's like, you know, I used to be connected to my Tiger VNC server by the SSH start the VNC server process. And then I can connect it with VNC. Although you can technically, you can force this on local hosts. You can force SSH4 in the Tiger VNC. Whereas with X2Go, it always uses SSH. X2Go client uses the live SSH library and then turn call some commands on the server like X2Go run command. And then turn launches X2Go, you know, display server called inXAgent, part of the extra package. But, you know, I don't have to tell novice users to run commands so they can start connected to desktop session. I just can just tell you the users, you know, connected to this host name default port 22 and select the desktop environment. Another example, a way to feature is, you know, the ability to select the session you want, like GNOME or KDE or XFC, or things like GNOME versus GNOME fallback, GNOME classic. Those are several X sessions. Whereas if you're long in local with a GDM or another display manager, you can select those X sessions. Session broker is, you know, it's not as easy to set up as simply it's longer than X2Go server, but session broker with X2Go, you can have your clients connect to the one session broker. From there, it'll refer them to, you know, which server to connect to, like, and by doing that, it can do load balancing and can have centralized configuration. So, like a lot of the things that you would normally send X2Go client can be used to what on X2Go broker, like, which desktop environments or about one of the good servers. Sharing out physical desktops that's, you know, that is a useful feature for, you know, remote desktop features to have, but the basic, it's not like the way I recommend to use either tagging or X2Go because the performance is always terrible. It basically is which, like, pixel escaping, like, render the local monitor and then try to send what's rendered to the VMC, and many people, like, through our work station right now, and GNOME has the screen sharing feature built in. I forget the package name, but it's using VMC, and, you know, it's good for remote tech support, but it's, you're never gonna get the good performance or other features like you can get with a real, you know, virtual VMC server, like, like, like, like the command I showed you. And similar to the fact that there's, you know, mentioned that X2Go and Spice have caching, well, that's because basically they use, you know, client-side rendering, I mean, X2Go, it's rather than completely disregarding the X11 protocol and using it in Xlibs to send X11 over, like, the SSH tunnel and, like, compressed in very different formats. So on one end of the tunnel is standard X11 clients, on the other end is an X server. So with X2Go client for Windows and Mac, we literally use an X server. But that approach of, you know, working with the X11 protocol to make it more efficient rather than disregarding it as why X2Go can have really good performance in terms of, you know, responses to users in low bandwidth. The one limitation though with X2Go and Xlibs is that it's, you know, an Xlibs was developed by NoMachine for NoMachine 1 and X1 through 3. They needed to open source, but they didn't open source the rest of their code. Is that right? And then with NoMachine and X4, I said what's going to go proprietary and go with a completely different architecture. So at this point, X2Go and other projects are maintaining in Xlibs. But until we, we're working on trying to port the latest desktop environments, like Decimars that require the composite extension or GLX 1.4 instead of 1.2, but that's going to be those, will be added in the coming months so they're worse like a year or so. Okay, now for the actual demos connected to this machine. Amazon EC2 instance right now. Well, you can't really see it, but it's an Amazon EC2. It's in the, you know, Northern Virginia. So I have that coming in in a script in my home directory. And this is actually a EC2 instance from a Fedora cloud image, an official Fedora cloud image that I installed actors on top of and enable the services. So just started the Tiger VNC service. It's just listening on port pole and one. So while there's, you probably use the standard remote desktop viewer, I'm using the official Tiger VNC client. And note that in VNC terminology, the viewer is the client. It's, I prefer the term client because it's what every other protocol uses to describe the client. With Amazon EC2, you have an internal IP and external IP. That's the correct one. So it's, it's wanting to know how to start applications on. There's some different logic for the policy kid in terms of these applications. I'll talk about that later, but here you go. Here is the Fedora workstation, no desktop and all its beauty, access remotely from Northern Virginia and the Amazon one pro cloud. There's the on-class servers. And the 3D rendering is done via LLVN pipe, the software renderer that's much faster than the old soft pipe. So even though, you know, the home shell is using geolux and deposits, there's enough CP power to keep up for it. But you will see, for example, that dragging windows is not smooth because VNC is, has to produce an entire window contents, although at least white, white, you know, the white area compresses extremely well in JPEG encoding, which is probably the following too. So if anyone wants me to show them anything specific on your work, there's also the Tiger VNC that always starts this by default. So we're going to configure settings. Sorry? What? What about video? Can you play video? Oh, I'll try it. I won't talk, later in this presentation, I will talk about video and how there's a really cool approach that the Arteca project has. Arteca has basically spent all five years to go, but for now I'll just show you this one. There's somebody giving me a glass of water. Wait a minute, it's right over there. I guess YouTube doesn't see many people accessing themselves in Amazon EC2's IP address space. They said, oh, you're robots, what the heck? The bacon looks nice, but often also kind of pancakes. That's one of the most creative captures I've seen. So let's try to make it easier for ourselves. We'll do like the standard size window, but I know that there will be no audio because that's not part of the UNC protocol itself. So there's, you know, small window size and sort of school window size. And thank you YouTube for using, you know, HTML5 and having free video formats. Oh, it's a close, it's like the BNC, like I said, well, I can show you that. You can use it in this session. One of the limitations that BNC has is, you know, unlike with Mode X11 or X2 Go, you cannot, you know, do remote applications. You cannot have, you know, the problem is you can run the mode with a local window decoration and everything. It's just the entire, you know, desktop window, but there are some features that they've added that some instances don't have, like I just resized the desktop session by changing the window size with my window manager here. One is that because I passed an argument with the BNC server call-auto-kill, once the GNOME session, you know, command quits, the BNC server quits, and we have a, you know, basically a clean log off. Next I'll show X2Go and then I'll tell you. So for those of you who don't know, Mate is the continuous of GNOME 2. I mean, it is intentionally a fork of the GNOME desktop environments, but starting with GNOME 2.32. And so, yeah, let me show you first how I connected with X2Go. Because you know, it doesn't have to run any commands, so. First of all, this is the standard X2Go client. There are two other X2Go clients, one that share code called, one is called Pyhoku GUI. It's a GUI also, but its focus is on having like a task bar tray icon and connecting the multiple of those servers at once. And it's Pyhoku CLI, which is a panel interface, but X2Go client, it's, you know, the most polished of the three X2Go clients, and it's optimized for like, you know, like clean client usage, where you have a, and single user usage, you know. So, so, you know, session name can be everywhere you want. You can change the icon, you can find a nice icon. Log in, that's, you know, it's Fedora Cloud Image, it's Fedora, and I'm using my SSH key, because X2 is SSH, so it has numerous, you know, SSH authentication options on. And, I'm not taking this to the desktop now, but for compatibility with old version, being accessed as a custom command sometimes, you can see the other type of sessions here too. And then there's compression options, I always leave it to default, PyQual and JPEG, you can change, instead to be like, use the whole display, but if you make it a window, you can always resize it later. Keyboard is optional, sound, printing, whatever sound you use, just pulse audio, and okay, I'll connect, using my SSH key. So, here's the, you know, the Mate desktop. And one of the first things I'm gonna notice is the beauty of, you know, client-side rendering. Just like that, I'm moving the window around. No, extremely slow drawing, it has to resend the entire image on the desktop, resizing, I like working pretty well. And, of course, that's no system longer, but, you know, generally, I don't have no maps of the Mate, anybody. You know, Mate's using a fork of metacity called Marco, so it's not using a Geolex or Composite, and that's why it's compatible with Xeo, but also makes things a little faster too, because, you know, having regular X11 windows that the Alexa server moves around is, you know, can be done very efficiently with protocols like NX. Anybody wanna see anything? Like, wanna see you two this time again? So, it's about similar to VNC. The VNC's probably a little bit better than the mode video, but, oh, I do have, I should have audio, actually. So, the audio is uncompressed. That's the limitation of false audio, so it's having trouble keeping up. But, on a LAN, you could, you know, the audio would work perfectly smoothly. I'm gonna show you a little quick, another example. So, I'm just gonna go, like, user share. So, there's a, you know, that is minimized and we store the window, it opens like that, because, again, it gets caching the, you know, the image contents of the window, okay? Anything else? One thing I would show is that one thing we've fixed recently is policy kit integration with X to go for desktop environment sessions. So, I open up a growth manager, all that kind of thing, manage the influences. Two instances, we have concept in, growth manager is tripping a policy kit window it's used like a year ago, the policy kit window would like fail to open and the growth manager would say, air cannot authenticate, but now it authenticate successfully. I don't have any damage running, but it's proving that, you know, policy integration is working much better now. Policy kit does, however, say that this is a remote session, so I'm gonna apply it in policies. Hence why when we launched in the home, it prompted us for like the password to like manage color devices, but if you logged in locally, policy would say, oh, you're local, I'm under warranty for your password to manage color devices. So, single applications, I'll just do the browser. So, there you go, I'm running Fox remotely on, I don't know if you can get a larger window, but full screen, if I go to, I actually identify this Dellware, there you go, it thinks I'm in Amazon data center, but it's, this window is actually coming over, you know, X2Go, it's a single app and you know, you can have, the same apps, you know, multiple open as the same application. The X11 protocol supports it and the XLIB supports that feature too. Of course, some features, some things like group manager would not work currently, I'll discuss that later, how we wanna, we're gonna X2Go on it, you know, hook it up like free desktop works, and it's so that those applications can work properly. See how it supports folder sharing? Yeah. Does that work for the application? Yeah, I mean, it mounts like a folder when there's like, 10 selects to go, Mike, I think. Let me test it out real quick. The anti-adjusted, full as after you've connected, I just haven't tested the pictures, working on it again, so I'll share my desktop folder, I'll just add files for later, and auto mounts, yeah, we'll wanna take a session. X2Go does have a desktop bindings also, so that you can see like, a nice icon on the desktop for the shared folder, but I think I forgot to install them. X2Go, there's the, while you have the X2Go when you open it, this is like, you know, part of the XLIB protocol, it's called an X-PodT, there's also an X2Go platform in the background, and why it's not showing up there. Oh, it's a hidden folder, that's why. I'll add some to my desktop, just for example, on the desktop, and it's basically using the SSH daemon on the panics to go applying to share the files, and then it's using SSH to press on the server to access them, and create new writes also, so files, there's free.bar just created, so yeah, I've got to install, you know, the Mate bindings package, but the function, if I had installed it, you would see an icon right here for the shared folder, and one of the X2 developers, Mike Gabriel's a Debi developer and a Mate developer, so he's, yeah, he's made sure that Mate works very well with X2Go, and then one last thing I want to show is, I'll show you how I'm doing in time, 15 minutes, okay, I won't show it, but so, you know, no, when GNOME was originally, through which it was developed, they had GNOME fallback mode, which was like, you know, actively maintained metacity window manager and GNOME panel for the shell, after with GNOME 3.8, GNOME classic mode is introduced, where it's using GNOME shell and mudder, and it provides a similar look and feel to the old GNOME fallback mode, slash GNOME 2, slash Mate, but it's using your frameworks, and however, there's, you know, you still have limited, still the issue with mudder window manager and GNOME shell that you're using, the GeoLux, Hardware Slug, or software 3D rendering all the time, and you're using them as extension. Jakob Solkowicz is a red-hatter who works on ARM for, ARM porting for Thinkwell or Fedora, both, and he's also a SIGWIN maintainer, and he's, he and I were both interested in getting GNOME flashback, continuous GNOME fallback mode, a package for Fedora. We currently have an Ecopra repo, there's the obstacle to getting it, entirely packaged in Fedora, because it's like five packages, so like Mate, GNOME panel, GNOME flashback is its own app, you can get GNOME, et cetera, is that the Anaconda OS installer uses Matassee, and they're currently only tested with Matassee 3.12, whereas we need Matassee 3.16, GNOME, and the R16, et cetera. So, there's some, we will continue to try to work on that, we just have to make sure we don't step on Anaconda's toes. With GNOME flashback on, we'll just fine, I'll just show it at the end, so I mentioned earlier that there's no better integration than just GNOME Flash Desktop, One of the issues with TigerVincy right now is that you have to rely on that script in the system or you're only going to have it in like com4.xcline script to choose which session you launch. Like you have to specify the GNOME session command or the MATE session command and you have to set some environment variables that normally define the accession files. The files, you know, users share such accession. I do have a proof of concept app where you can lock, type whole thing into the app and then the app lets you pick the accession launch. So if I just go to a computer and then you hear a share accession, you just see what I'm talking about at the time. And right now there is a commercial, you know, we have a desktop version called Thinlink that's using TigerVincy but they won't miss a stage like session handling and they have a similar approach of managing sessions. So these are desktop files actually that you can share accession with seriously. Can you guys see the text? So here's a dot desktop file that says, you know, launch this command for this accession called GNOME flashback and make sure that you will install and set the environment variable, couple of environment variables based on this value. But this isn't, you know, it can be complicated but the problem is right now we're relying on those scripts now under slash ATC or use of home directories and those often get out of sync with the necessary environment variables, for example. Policy kit, I think there's still some many issues like YUMEX seems to be all authentic and I'm not sure why someone gave us some, you know, to look into but personally at home, I'm using threat manager all the time and I'm glad that that's those ball-scale things to work in. Another issue is that, so it actually supports a single app mode and on the Linux desktops, you go to the desktops EDC slash XTG slash auto starts, you'll see all these automatically started apps. So, for example, whenever I try to run, you know, Firefuck Thunderbird over XTGO, I need the GNOME keeping daemon running so that I can unlock my GPG key. But if I do a small app, the GNOME keeping daemon starts, we won't be able to keeping daemon start. So, I would like to work with free desktop.org, aka XTG developer standards, like these apps should be started automatically when you bring a small app here because if I just do like right now, you'll see right now it's saying that don't only start the GNOME keeping daemon in GNOME, Unity, and Latte, it will never attempt to start it in a single app session like XTGO has. All right, so, the next part of my talk is, it's rather, you know, I started using XTGO artwork because, you know, I had security requirements to meet and XTGO was the easiest solution for me to access servers in our lab from our Windows desktops. And none of this here is specific to my job at Lockheed Wine, but it's some of the issues you'll commonly see. I literally met a developer, a web developer at the local tech meetup. The application he develops is on well server and it's a web application for financial companies. But he literally told me that. He said that IT security is stupid and you don't understand Linux. Therefore, they won't let me run Linux on my laptop. And in other companies, you have a lot of other issues just running Linux on your laptop and desktop. Like, corporate IT won't support you with like failing hardware if you have Linux on your laptop or desktop. So, there are standard wins and wins in corporate IT via the hardware, but they're not gonna maintain it if you're running your own desktop OS. And in addition to having issues with people applying to run native Windows app, native apps that require Windows, even many web apps developed internally by companies that require an explorer. It's like we build on brand new modern web frameworks. They'll be one or two bucks with Firefox and they'll purposely never fix them. They'll go on a minimize cost by only supporting IE. So, I know lots of people with their heart and souls into Fidora Workstation, but there's lots of cool technology coming out that are running Fidora Workstation, I'm sure, but in many large enterprises, there's a minefield of problems trying to run Linux on your laptop or desktop. It's just remote into a server or a cloud instance and access your Linux desktop that way. So, the future of remote testing in Fidora, so I already mentioned don't flashback at the Alcatraz software, so I'm working on coding. I've mentioned integration bugs. I mentioned that the accession launcher. And I do also want to point out that, well, Tiger VNC and other VNC solutions and X2Go can show an existing physical X server that performance is always much worse. Like many people, those like in Fidora Workstation are now able to go to sharing, screen sharing, and turn it on. But, you know, well, all this works and can be easy to stop. You will never get the same performance if you run your own Tiger VNC, you know, process separately. And the future, well, so the Oracle Project is basically a spin-off of X2Go that's, they develop external features. This is probably the thing that will shock people the most 4K video playback. I mean, obviously my screen's not 4K, but it's via the telekinesis framework for X2Go, although it's not going to actually really include an extra something in the project instead. Well, the idea behind telekinesis is, you know, typically with remote desktop, it's just like Microsoft RDP. In order to comment video, you know, the server will decode it in CPU or GPU, like, you know, the visual flash player or in-player, whatever. Then re-encoded for RDP. And then the RDP client has to decode it and then show it on the screen. With telekinesis, you're just letting the, you're just like forwarding the video without re-encoding it from server to client. And it's fairly easy to do that because SSH opens up lots of options. Like, you can use Unix sockets to do, you know, secure, like, permission-based communication between server and client, and only the proper user accounts from server and client. So it's, you know, how much balance is required, the price, however much the 4K video was encoded with, plus it's a tiny bit more of SSH overhead. And, you know, and you can do things like buffering too, because it's like the YouTube video because it's forwarding video over SSH sockets. And then the other, this is my final slide. So, you know, Wayland's open up a lot of possibilities. I'm very excited by LibWayland architecture because part of the Wayland architecture is that, you know, the wayland itself is LibWayland. The compositor slash moving manager links against LibWayland to become display server also. And that same process is also the screen lock. So it wouldn't be possible to do things like what Microsoft RDP does where you'll log into your desktop locally, you lock the screen, you reconnect via, you know, via some desktop client from another computer, and make sure the screen is locked locally and then you're, but you're still using your set, the same testing sort of locally. By locking the screen, you can have the same performance, you know, performance of not doing the pixel scraping, like you share the local screen on. I mean, unfortunately XTO would not be part of Wayland because, you know, XTO, you know, in X-Libs works, it compresses and caches and eliminates the bound trips to X-Libs protocol and based on X-Libs protocol, both ends, it's, but I'm just hoping that in terms of tool kits and possibly display servers themselves or the pros themselves, you can influence some sort of client-side viewing solution. There was already some work on like RDP as up in, but like the guy who was working on RDP hasn't open sourced it yet. It's called, it was improving free RDP client and the free RDP as server based on the free RDP client code. So that's it for my talk. Any questions? Yeah, right. I mean, that's a, like all the browsers now support WebSockets, so it's not a problem. I mean, okay, gotcha. I mean, so a lot of people ask this, does XTO client have an HTML5 version? Well, XTO client, you know, it's expecting to have an extra, you know, an extra for all the clients and that's what makes it so efficient is it's, you know, doing X-Libs commands rather than doing, or small trade protocol. So I think XTO client needs to know for client, but it's efficient for that reason. And also there have been some attempts to develop an HTML5 X server. I think it's like, you know, AlphaStates X server on YouTube, but even then, you know, on GitHub, but even then you'd be like, you know, you'd be seeing X-Libs commands to the browser and then browser has to talk to the X server. So it would never be the same efficiency. Any other questions? Comments? So is anybody here, like a, like a, like a software developer on Nome or through our workstation? I mean, I think I'm tired of that. You know, I've, you know, there's, there's one guy and a couple of people who contributed to, you know, Nome Flashback, the Outcraft software, which I have worked on impacting it for, you know, Fedora and with REL 7.2, which will have NomeFeeder 14, we hope to have an EPL for that. But, you know, there's lots of, like, lots of people in the Nome project think that we're trying to, like, fork the Nome desktop. We're not. We just, we want to maintain this alternative shell so that we can have better performance on things like, you know, low power ARM hardware and, you know, mode desktop solutions like BNC and X2Go. Like for, like we submitted a couple of patches to Nome Settings Damage and there were certainly two patches and they rejected them saying you should fork Nome Settings Damage and we don't want to do that. I know Unity, Canonical, Ubuntu had 61 patches of Nome to Nome Sets and they, setting Danes before they forked Nome Settings Danes to Unity Settings Danes. That's not what we're trying to do. We do want to work with upstream projects like the rest of Nome. It's over the right when you go. I'm sorry? That's certainly in the right way to hope. Right. Right, I mean, original Canonical said, Unity is an alternate shell for Nome and now they've backtracked on that and just making it into its own desktop environment. We don't have the, you know, the resources nor the error against to do that. Any other questions, comments? Okay. Yeah, you'll see me on there, see all the time. I've also been hanging out in Fedora Cloud recently. I've been working on joining X2Go with either the Mate desktop or, you know, single apps in a Docker container. I'm still trying to fix some bugs but the Fedora Docker files for X2Go and X2Go Minimal should be included in the Fedora Docker files. So one thing I forgot to mention over there is but X2Go and other solutions to run containers that can use system D to start SSH server, for example, make SSH server running on like, you know, port 50,000 on the host or native port, for example. And I think I would point out, you know, in terms of like mode desktop, it's like it belong to Cloud SIG, the desktop, the workstation SIG or the server SIG because all three are valid use cases for X2Go and other web desktop solutions. But in terms of integration, most of the integration bug to have to fix it in terms of, you know, the desktop SIG. Thank you. Well, if you're still on time, I could show off a known flashback real quick. Okay, I didn't know if you had to close down and you knew what. Yeah, which system.