 Hello everyone, can you hear me okay? Can you hear me? Yes. I'm Benjamin Weather and I'm going to talk to you about one check-in stall in 18th September. Let's talk about what one check-in stall is and how we're using it within the check-in system community since 10.3. And how you can use it to make music for users to install stuff there. And then if you're not tired, you might go into some of the problems and how you can make things even easier. It was a long time ago. So, what is one check-in stall? For music really it's not actually installing stuff over one bit. The name stuck not entirely sure in here, not really. Amazon. But they're much better, I think, than some object users. It's not as bad as it is as Amazon. But what it is doing is trying to remove some of the problems that users are encountering and trying to use the package language system. So, problems. User visits the website of some software such as a camera book or pigeon or something like that. And then when they might try to find out how to install it, then they might get a lot of instructions as in terms of the methods that they're doing such as the software repositories. And then they'd have to follow some steps to install the software in order to make it easier. And then in order to do that, they might have to use some complicated interfaces. This is one that we had in previous versions of VocalSutor for adding your package repository. And the user would have to know to split up URL into the server name and the directory on server, which a lot of people had problems with. And it's not just one step. We might have to add the package repository for the particular project. And that's like while metadata from that repository is downloaded. And then start the package manager up to look for the packages. Select them to be installed. And then wait again while they're actually installed. So, things we could make easier. As I mentioned, the user visits the website and then decides they want to install it. Or, you might get a set of physical CD media. Or independent software vendors such as the gamelike and autonomous. They tend to use their own custom installers that don't want to go back with the system package management. Do you know if there's a way they can go and install it that's both easy to use and integrated with the system package management. And also how to's in the tutorials. For example, how to get full multimedia support in a system. And how to get this hard work to work. In the past, there's been voted a lot by having to include explanations of adding your repository and installing these packages and then performing these configuration steps. So, the solution is to automate the addition of the required repositories and all of the packages that are required to install the software. So, on there, that is 10.3. Right, click a link on a web page like this. And then get a wizard like this which will explain what the software is that's going to be installed. And it will go ahead and install all of these packages and use these package repositories to do it. Because beforehand, these are a package manager management system. And then tick all of these boxes to install the software. And if you carry on, there are a lot of great machines in install. So, how does it work? It's a simple as well description file which just specifies the repositories of the software and the packages that are included in this file of software. There's a special line type and there's the handbook installed on the system which understands this as well file. And it will perverse it into package management instructions. And you can, unfortunately, use the data postal which is made up of most of the browsers to get around the fact that if you don't have access to the web server, say, for example, if you're not posting an email, you don't have access to your total of one server that you've already submitted. So, this way it works well with the browsers, it doesn't provide modifications to the browser to all this configuration. Some other ways could have been done. And you could have RPMs that can install the repository on the system. But you need a package and knowledge to convey those when it becomes a lot of distributions. And I think Ubuntu has been at your ritual, it's people on the list version of Ubuntu. It can install packages from the main system of the system. That requires more than services modifications to the browser. And you're a bit more limited to what you're interested in. So, have we used most obvious ways on blogs to develop this buggy about new software that features that out into software. You now add links to the blogs that the user is going to be able to look at it. I don't think anyone's using it on media. So, software catalogs pack-bound is one of the most popular third-party repositories I've consumed. They include a strict codec and lots of other newer software in the shape of the distribution. And on their website, there are a lot of package details that don't have Ubuntu and still links. So, if the user's having boots, they add it to the pack-bound repository to their system. So, they can install them easily. Also, build services work in a place that makes use of it. And they can add that in such a way. The traditional version of trees is being a particular problem in open-source because they do other distributions. We do have an awful lot, I think maybe up to a thousand, because a tree is available for each version. Whereas other distributions don't have just one big one. So, it's also used for the yes community repositories module. This is a new feature in 1.3. And it is becoming the most popular repository to put these to my watch with a lot of descriptions. And they can choose which ones to do and which ones to do. This will download a file in the same format in the server to get this list of repositories. Great. So, if you've got some software and you want to use the one-second store to make it all simple, use the system and install it. First thing is to package it for the entire distribution. So, create some of the RPMs for everything. Then put those packages in a package repository so that the package manager knows better than we can. Find what some of those are required in order to use the one-second store. And then you can create a YMV file. This is the XML description that we've mentioned. And then you need some distribution. The best way to do it is to access it. So, the build service can help with all of these steps. You can enable packages for all the distributions and you can host the repositories if it's great software. It can automatically generate the one-second store links, but you don't have as much flexibility as creating yourself. So, XML format. It's relatively simple. You have a group of packages which represents the software bundle that the user wants to install. You can enable them on a base or on a YMV file later. Sorry, can you speak a little bit though? Sorry. You have a group of software that represents this bundle of software the user wants to install. So it might be local packages that go by an hour. For each of these groups, you specify the name of the software. So it might be on the way. Give it a summary and a description. This will be displayed in the user interface of the wizard so that the user knows what's going on. And then you have to specify a list of repositories that you require to contain your software. And a list of actual software packages. For each repository, you specify a name and a summary and a description. Which will be displayed in the user interface. And the actual base URL for the repository. You can specify some word information here. This is about the mineral that's arrived there. Name, season. And then the package actually will be able to have the repositories from those URLs. Each software, specify a name, which is exactly the same as the name that will be displayed in your package manager. This is how it names what to install. Add a summary and a description, which will again be displayed in the user interface. The reason you can't get the description information from the metadata that already exists, is that this is displayed before any package management operations have performed. So, it is the reason you need to decide whether this is actually what they want to do before they are going to be commissioned to actually install the software. A couple of other things to mention. You can translate the strings. So, if you want to support users in different languages, you can specify the same description again and a summary in a different language. And the handle will choose the one that addresses language for the user. I mentioned that more than one group. So, you can specify instructions for differences between 10.2 and 10.3. The handle will pick the group that matches the install system. So, if you have different packages on different ones, all in different repositories anyway, you can use that. And just have one in your link, which will work for the user in different versions. And if you don't specify one, that will be used in other languages. Is there a specification for how to use the string in a disk version? Yeah, it's using the version that, say, the evidence is, once using the one, the version string, but the answer to say, yeah, it exposes the instance in the etc. So, it's a release. If you want to differ with a lot of distribution, then there's a person that will be able to decide what you want to use for distribution. Yeah, so you may be thinking, I don't use that in too many ways, is this, but you could potentially add the implementer handler for this XML file all the way for your distribution and convert to a quicker package management construction to your distribution. And if you find that there's additional information required for the package management in your distribution, that isn't what they can see, then it is possible to tweak the XML schema because, in certain ways, they won't break back. As far as sorting out the distribution, another thing that might be nice is using some other package management system such as CLIP for users who don't have their boot access on the system. Is it if you do have support right now for Fedora on your package trip? No, package kits is a difficult one. They're not sure if they have a concept over a puzzle train. You're not seeing anything in package kits then in case you can add and remove the puzzle trains. You can implement potentially just installing packages that are already available in the system of the puzzle trains. You could actually use Fedora a lot, so I'm curious more about Fedora itself. I'm just going to use a package kit to help you use that. It would be a matter of time to do it, but if you need to support it at the young or simple level, then you have to do it obviously through... Well, you can package it onto anything, but if you don't have support for adding the puzzle trains, you can't really package onto them. You could with package kit and something else. I think we can do a local and soft file stuff. It's probably ready to be done at least. Nothing else. Am I right in thinking this has been back ported to 10.2 but not previous? Would it be hard to back port it to 10.1 or SLE 10? It probably wouldn't have been actually traded at the university of 10.1 in stores. Talks to you after? Yeah, okay. I am holding the time. Fedora problems had the one... The handler included with 10.3. It doesn't understand installing only any of the versions available. It's a software, so operating SLE all the kd. Also, that moment just the best match for the package gets installed. It was chosen by the seller. A lot of people would like it to be installed from the specific repositories that are named in the X file file. This isn't normally a problem, but things like analogs may be available in three or four repositories on every season. Then it's said that we may not get the version that they actually want it. Some people are requested to use it in place. The build service can generate the version. Another problem, possibly, is that it's not too easy to install software. It's very easy now to install beta software with a system or even malicious software. They have the protection of the package management system, so at least we have to decide whether they want to trust the key that the software is signed with that to most users and other messages such as that. Well, warnings such as that, whether they do not mean it or not. They wouldn't know how to go and check it and find out whether it's trustable. But there are people working on trustability service and they're already waiting for the build service. So it's nice to look up on the build service or whatever, how much the package is trusted, and whether this package is high quality and so on. You might be able to display something more meaningful to the user. Another related issue I want to talk about. This makes it very easy to found it, so if you go to a project homepage, you can answer it very easily. It doesn't solve the problem of finding the software the user wants if they don't know that they need it. For example, Codic installation. It's a bit better in 10.3 if they use anilog or possibly want a big amount of players. It will ask them if they want to install mp3 support and it will guide them eventually to one of these point of the stores and it will install community package. But it's still not as easy as it should be and it's all those things that aren't mp3. So it would be nice if the system could find software more automatically. Hopefully, yeah, we're going to have a command not found feature which will tell them if they type a command and it's not in the store. We're going to try to install this package to fix it. It would be nice to also be able to say that even with dialogue, when you plug in new hardware you can look up what gliders could be installed to fix it. Both in the local package management caching must be with website schools as well. And it has a list of software that could install it. It's important for them. And yeah, Windows does this already. It's potentially legal if she's... There's lots of software that it wouldn't be any problem with, but the sound of mp3 support by video drivers is potentially legal if she doesn't make a client with it. Or the number of websites up there. And even with problematic software that are way around it. Community repositories, YAS module. Also this is the new repositories that we've found that's officially supported by Windows. And also we've had a host of this that provides us with that information about the host to get themselves. So yeah, that's something really interesting to work as in the future. My pet peeve with one console is that. It's being installed at UI when you're installing different packages. You just get lots of alert boxes coming up and up and up again and sometimes lots of focus ceiling. Is that being worked upon or is it a scope? It's just using the standard YAS progress reporting. And that's already being fixed in the 11.0. It'll just give you one focus plan. Gollos to other distributions. It doesn't have a package key. Is it just a number of changes? But there's no support for package keys at the moment. Yeah, but it's dropped. There is no package key, so it's just using it straight off. But it couldn't be backed onto anything that's supported by package management operations. So you're hunting young. Can you have a wine piece with just the repository and no packages? Yes, so that's what they came to. More information on the website. Or you can come on me for it.