Bestand wählen
Merken

AppStream & Listaller

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
this is the revolver technical the this is a period of the the wall so
i'm parameters so I will talk about gas stream and the solid today resource and boring stuff about me when I am a pact to developers Debian developers contributed KDE and Gnome and
and the crucial maintain and maintain upstream together with over to choose but I also started the time projects uh last year together with other people of the yeah but it's not more important than the what's there will talk about the 1st of all I will identify the problem but what what are we going to fix with apps stream and scholar
and then I will talk about are stream then about this tolerance will probe the make this a bit quicker so we have some time for discussion and the and and and and then at the end I will draw a conclusion about the both projects the 1st of all of regional what is wrong with application management to understand why we need that upstream and 1st the on the content of the distribution suffer repositories this usually displayed in form of packages but this is not very user-friendly because every non-technical user finds tossed packages from where some have names by which he doesn't let you should and understand and are mostly the way there are there so many packages and what development packages for and if you just want to install Firefox for example you bitch easily get lost in the management of also the existence of a suicide distribution specific and basically every new distribution of power would duplicate the effort of creating a supper center was center is an application which displays applications the set of packages so compared to an app store if you want also the applications are not very well represented to user because right now we have only the deaths of file metadata source and Nikon which is usually associated with that of bad no additional data about for example a localization is often missing not and at the end we do very well with that that's the other distributions don't have that and there are no screenshots or Basque region and there are lots of inconsistencies between distributions because every distributor writes its own description for every single application every single package I and also there's no interaction of possible from side of the user with the use of a center for example it would be really cool if the user could give a rating a review for a new software or uh if you could just reported back within the office and user interface because if you just try the application and it didn't work for him and he he found a black them it's a more natural way to report it so that yeah this is about the application management but there's also another issue which is application distribution on multiple Linux distributions of the current because of situation is that we have to keep personal package arches and you want to have the will to wall and repositories for every other distribution to get new software but um for third-party applications and these are really insecure because they can break distribution upwards of half of third-party on PP and therefore repository enables which of packages and application differently than the distributor also packages that thing but also slightly different and then the the the the user wants to operate a system he will might easily break it a conflicting packages also their scripts inside these factors which are executed by roots so so with her permission so is if you install the application you can be assured that it wants of 1 execute malicious stuff during installation before we can start box that but also having a few cases not very user-friendly right now we want to test some work on that but it's still not perfect but also because can be as a distribution specific if I want to package the third-party application I have to create 1 repository for every distribution and the RPM worried about what more almost 1 RPM Package where it is used by and also that the complex structure because there are made for of a package distribution and not for this edition of applications which is what you're aiming for and that's another solution to install third-party stuff which is binary installers but these don't integrate well they are in a user-friendly and they're execute execute lots of stuff with her permission if they need to be executed as root so we need a simple integer cross desktop of insulation solutions for third-party applications which are not yet packets with the distributor who might not have time to backward everything so uh this S . 2 independent projects 1 is apt stream which is an a cooking which many contains course this those specifications for bowling suffered center and a center like applications of which includes a lot of metadata database us back 40 signatures axes to get information about what
applications are available in the package databases I also there and it includes some uh option new metadata for upstream projects to ship so we can enhance of the available data do represent an application in center of they also offer the attractive features I spoke about before I was taller and the other and generates a process application packages it's and it's a slap developers to make applications and available for many reasons and that it contains also a metadata specification to describe dependencies on the distribution it is an optional component so you if you don't want to you don't have to ship was tolerant of and and requires packets queued so if you solid should uh make sure that packets excuse is available the so 1st talk about Astrium acid was 1st created in 2011 by these people from Fedora adapt open Susan Mikhail going to KB and some for death of developers and yet they sat down together and discuss how we could collaborate with which and how we could do a collaboration in order to create a really good looking so percentages available amount distributions so not every distribution has to read do all the work began to make you suffer center happen good and so this resulted in the Aston concept which um stores with the so-called composed of a verb which scans all the packages which are available in the repositories extracts specify files icon's and app data will come to update later and then generates a large XML file containing all this metadata and all translations which gets compressed and shipped to the client together with and I compact and the client then of aggregated data from multiple sources 1 of the the stuff which came from the composer uh but also stuff from a speech observer a form of feedback so rather than other possible sources to display the application to the user in more detail the Odessa files text files are of converted into avian database for a partition of features that we see standing on and faster access to the to the information yeah so the g of Astrium is the so upstream XML which contains all information about the application which is the name of a unique application ideas so we can identify the system and we can ensure that there are no duplicate applications which is another the deaths following a summary of some description I can often data for the group which is for example no more a KDE so we can filter applications if you don't want no applications you can say no go or if don't want Katie and a of KDE and you can also create different departments for that so Mozilla is also predict group by categories which is I'm available in identifies already mine types so you will be able to to credit to create a cry reform in an application which is able to open or to process and and tired which you don't know which is also a useful feature for users to get a file which they can't open so that would just have to suffer center how can you in that and keywords self pretty self-explanatory and speech of references and description so we have every application can have either the screenshots which are also which also contain a small description on what the user can see on the image is a contains a few other things which are less important there are a few other issues with that which experience after writing the upstream XML which is a that's what a so it'll also allows localization but uh their futures we noticed after it's and surprisingly some people don't like XML of which was a bit embarrassing for me because I 1st created that XML together with the others and stuff and then I want to implement that and that reinvents aggregate people told me that x analyst nowhere in the infrastructure and spend their Darien prefer stumble for that instead of axonal of so we created a new and Aditi an extension project called at level which will enhance of not only the which enhance metadata and deviance general it would contain the upstream data but also data about additional components like like Codex themes or whatever you want to find the packaging database and so it's a little bit more than just the extra maximal the other problem is that when to get data from the the distributions don't have it in their repositories already so that it can extract it from the death of file or if it is not already available in the description and the 2nd problem is addressed by a so called AP data at data supports a small XML files which are ship with us predicts that they are a subset of the estimates smell specification and they contain your else to server service which contain the screenshots so this is where we get screenshots 4 of Freud upstream from the umbilical upstream so uh and we have 1 description which has a localization which you don't have to redo it again and again and every distribution of and that no requirement to implement Asterion because of what we had some confusion about because the Torres said that they would drop and every applications doesn't have enough data file from their suffer center from known suffer but this is a distributors policies so abstinence perfectly fine without having enough data file it will just not display that much data about the application so it is desirable to have data file in upstream projects so if you should agree application but please provide an update a file for it did the act of of I the external content sigh screenshots uh if it is interpreted as like a said before it is some is going to be to be broken soon because that since the amount of the upstream server might go down or than might be malicious that even in there and then I change it afterward something so that the distributor on creating of the upstream XML Park will will download this scattered on its own server and modifies is necessary so of the distributor still less control over that that's the not 1st in the information lays on announced a service and some statistics about the and the operator of yeah there's not much to say it but it was really interesting is that we know already ships over 50 per cent of its applications with outdated because this made mandatory to have that data about well there's only 1 KDD applications with apparatus so I give you really have to catch up there I am drafting of something which might uh enable having easy upstream on Acadia yes well spectrum with some issues with translation but I'm pretty sure that we will fix that in time but for access the highest assumed to rectify the right people so we can talk about adopting updated they wanted the I should also mention another issue before which is the that we have to file formats 1 Jamil formant and 1 axonal formant and that's also because of going to was 1st entered the differently and a bunch of format which also contains information which we want to have for supper centers so the we also have to have to have full what solutions for creation of this address it's getting a little bit more complex at least during the transition phase that not everything is as it is in the upstream maximal formant around yeah so we invented the stream which is like a common library every news office and you can use
to access this this information we and this will of avoid that every service and that's of reports the XML file or unidentified or the would Japanese-style style information or has to download the speeches from Essen take screenshots there'll done what the people of the screenshots from there are and stuff also consider a fallback solution of the distribution of force 1 so this absence that generates a set in database which contains all the data points of a centrist taxes so the sub center would basically only CU-Move database and not the raw data was generated from and also live upstream ALS accessing the database by IID object interface because many people were afraid of using the native 2nd interface and the object elements creating bindings for almost any programming language out there this is how it looks like in the end and we have an upstream black in which runs inside the containment so it will listen to refresh actions and on the pact demon and if it receives the refresh action it will check if the action that is up to date and if it is not all the data from would Japan's solved and that in that 11 off last year XML which all of distributions ship and um that generate the 2nd database as you can see it also requires of information from different sources and the clients if they want you can access the 2nd database directly so they don't have to use the substrate the so I won't ask adoption in the middle but don't suffer 0 was created last year with optional of is the only really axiom FileApps incompatible suffer sent out there but it's a full support in Fedora and it's pounding on a on open Sousa as far as I know and and at the end of it's like but the debt level implementation which seats just need some more time but on KDE we have copper which permits the KDE actually front end with its own aims to be both a package manager and an application area so to combine both words after a has some initial support for upstream by live upstream but that need some more work because it's not working as well as the as the 1 it's uh so there are some issues and editor inclusion into KB projects something we discovered now there and I also heard about a possible part of the kappa of and stuff to packets scared and upstream so I think there was a lot of people working on that at least 1 Summer of Code project if I remember correctly and the yeah there on the that distribution size deviance already talked about going to have the Apostolic which came before upstream so it's not the not invented here syndrome this time around it's that will likely migrated deviance solution but we'll see what happens the Fedora is Fedora they were pioneers on the upstream front and ship that stuff 1st and the world it opens user and has so has working out you maximal that's I'm not sure if it's used productively and last and they wrestle hiring out some issues in the generation process and so far unfortunately only know has a sulfur center which supports everything from that same specification so this is about the application management part so now come to the when I come to the application distribution part which is the sellers domain so 1st some some by going about the project was started in 2008 as an experiment it was 1st never intended to be used productively anywhere but it had covered some features of packet gets there and upstream that time it has its own package management abstraction layer just acts could have and it has some some metadata extension does like after has much simpler because of God's better respected in 2009 for a deeper talking for communication with the native actually enter but then later it merged with what a package and if you remember all too packages was started way way before this was created and ships binary packages within a development slowed down at a later merged with the solid and we adopt that as some of its features and some of the projects which were created inside a package of Asterion Watson sort in 2011 so this part of the cell was also so we re all that the income from scratch I would to which of all most of you and now it only focuses on the distribution part of which starts in my opinion really well known follows goals integration with the system so we don't want that the user has a number of insulation go we have to and to install an application while there already is a user interface with the which is the sulfur center on the packet gets application installer of also know we don't want a 2nd data so if you update a system that is active in the system it should be on that it should be 1 of a 1 time things so he has 1 up there to way up this ball applications and not the 2nd 1 like in windows and also with all of this all publication should look like native applications but they should not be treated as a native applications the so all of their own in the sandbox have different privileges and but because they come from a potentially malicious source and we're not checked by the distinctive obviously because this year and the compatible but it should be simple because we don't aim for a catch-all solution with the systolic it's should install applications but the applications but that's it is not insult stuff like in in system or uh install a tool kit for everything that has which has to reverse dependencies but and because of this because this is the domain the native distribution package management is working in and when it makes sense to have it while the solar is designed for only for applications and it should be secure which means we have security of acidity database signatures and sand box which is currently under construction and we also should some developer tools which are basically helpless to to kind generate statically linked applications in order to make it repeatable or to do some automatic QA like taking all of of the validity of all 5 formants used in application the so the basic concept of the solid that it is the Packers get pocket which explains the integration part because every application which talks to scared will also automatically talk to the solid because it's it's as a inside active and some and captures all those followed packages was a passion passing through and processes them before they reach the native back at it as and actually it if you don't know about exited the patient has different back and the 2 of them to talk to the nature of the to the native distribution package manager and as source of just sits in between that the well therefore as a set every exercise is able to install solar packages but also and this is the connection to upstream a solid job some XML into the XML pool of last which results in a really great representation of everyday solar cell application in this office center so if the if the axles present it will be automatically processed by the by the tool which generates all 2nd it is for this office and and uh therefore the suffer center and know that you will not be the solid package through the application information and displayed to the user so you can be uninstall it
for example also this study uses a superset of the update suspect as source for application metadata so aerosol package will also include an annotator XML file so if we are going to sell the application we can already show information about what the user is currently solid a yeah well estimate imagine a consists of many tools to achieve that is a testament to the tools which a man manage for example a database of trusted keys so that it's you to consent which sources of trust and the other ones are untrusted activities that's a key manually they exist some some in from the sun developer tools to generate portable applications link against an old you'll see version for example or the reversible static linking around it also contains some tools to track dependencies and to visualize them it also contains the obvious packaging tools like a like EKG at the end which is the idea a solid package generators and the smaller kids QA tool which for example validate XML files which were shipped with the application or check if the fights technically is OK so and this be just don't have to worry much about costumes creating crappy packages because of the packaging itself is then simple the problem can't do much wrong with doing that and the other stuff is checked by the QA to automatically so we can already see common mistakes it's just like Lintean but less powerful options of are so if you create a package is a utopia packets creation so really the the simplest way to create Napster package of everything uh subpackage if everything was passed intended is that you have to write your app data description then 20 small options file which basically just defines if you want to make a scholar automatically detect all dependencies in your source tree in and which go version of the cell factors you want to create and then you need to ratify the actors thing for the new package but was solid to decide where solid application based on distributed policy on that that's the and you need to create the dinner directory structure for the application by itself using some of several some placeholders for the final path with the cell which is so I will insert during installation they can already below the packets because they're so small to included which was detectable system and how old insulation part of and it's uh currently doesn't work with many both systems so that you can still give us all the required information to build of the application or you can just pointed to the binary stuff and package that so you don't have to build a package during the do with application during the packet creation process and yeah this is a script that only the packages to be designed so that any factorization which is not signed will be rejected for insulation and less but this future has slowed the port policy to all dangerous action selection so we can expect that has the texture signature and again during installation was sold process all files installed by by and by the package and also to some modifications if necessary for example adjust paths and of files or of well do almost anything the distributor of finds useful in order to integrate that applications right now it's really only of take a look at testify as if they're saying and if they are in the education so on and so I am I not talk about dependencies solely with which are my confuse you are also know presentation with all the cats and because the media says the style should be simple and depends solely as 1 of the most complex things at least 4 if you enter the course distribution of of small so why has this depends solving if it's so painful to it and what mainly if we statically everything into into 1 binary we not only use more memory well because we have to all that all that stuff but also the distributor for example pushes an update each onto toolkit which poses the critical as security-sensitive back but it will not breach the 3rd party application which also ships that toolkit so it it is a potential security risk which we don't have so it makes some sense to have at least a base level of dependency solving for major dependencies by example of system components like to it what toolkits so then the are the main problems here is that we want to simplify that that the distributor does not might not provide enough metadata follows follow to find the dependency in the native of distributed database so in that case we can own application for dependency is missing and also what to do if the ABI for dependencies of stable for example if they decide to yet to to make some changes without something that's so name and uh well the application just can use it and and also what do we do if we absolutely require an old library version well the answer to the last question is obviously static lingered for the ADI stuff but there is no real solution we simply rely on upstream projects to on 2 and yet to take care of that and don't break the IBI unnecessary necessary the in order to enhance the metadata about uh global dependencies and distribution repositories we created the circle component files which simplify it and dependency solving would go a bit faster here and components are for example to I already explained that 1 and OR 2 solar cell application and the so only the basic stuff so all small dependencies which don't a component file and ordered available might have a decision has to be that of you included directly in the package that you don't necessarily need to be statically you can reflect the thing is a library of the solid take care that the application confided later after it was installed in the yeah and last but not least but this I can in theory also satisfies dependencies by downloading a set including package from 3rd party sources which might not be wondered by the system administrator so you can disable that as I can also checks depend crack pipeline or the general memory in order to satisfy tendencies of stuff for example in so is component information and what information is can be compared to a non non-developer package config file it's describes all of them the component version and all the public interface of this provides which are for example libraries or binaries but also stuff like I consensus on and off the data which is used by and by other applications applications other than the product itself so everything which is shared should be defined here and later we can than doing a packet creation process map of the dependencies we found in the applications to indicate to a component tell the user well you are shipping stuff you which doesn't have a good representation upon representation on the distribution so you might want to to statically linked that to included there and we can also check on which distributions of the application rerun because we have this information we know which distribution that so we can run a check like yours do the border in the U. dependency is too high to run on Ubuntu 14 . 0 4 of something I yeah it also
allows did not make a dependency detection dynamic person detection sorry which means that if we took the component definition with the package itself we can only use and of the target system sack of the dependencies solid without having to carry the energy package manager the it while and I already talked about that as I said before we should that that scan tool which is able to perform that task and to not only detect dependencies of binaries but also Python models and almost anything leadership with the application of the I'm sorry for being faster here but there are some discussion time what you want to know the future of the solid is to 1st finalize and formalizes the specifications because some stuff does not have the documentation yet although it is getting better and then we want a formal standard for for creating some packages so others can review that and spell check if it's as if it satisfies their needs and also it's will result that in in a stable of accented so we won't break up 1 break packaging something for so as in projects at least some such component definitions by default so that we are able to send the query to be native distribution database for example is the decay of 3 the bigger than 3 . 12 inversion and this is also this is not only helpful for the salary but also for projects which compare versions available in multiple distributions because we can just check for a component different files and then map them between distributions and stuff detect which distribution has which which it also will allow us to get rid of horrible stuff like of evolve developers did with steam they should small costing that's which is heavily dependent on Daddy and for detecting dependencies if we have component definitions of steam could just to have a package get basis of file which is able to find find all the stuff they own the need for search for party sources in the distributions of repositories in across distribution away without being dependent on going to Italian courses scholarship should go into more distributions is already shipping sound but span the coverage is not really high so far and and maybe in the and have a way for fellow party developers target all these distributions with 1 Apr package because it's really desirable for it offered them up stupid dates to have 1 thing that users can download or access from a single source and yet they haven't lent creating 4 of 1 packet strongly defectors for every decision are coming together some of some weird binary stuff about all projects while I skip that because you've ordered tests recently and we like in countries just upstream and to install but because they're Covey very few people working on it and it would be great to get more feedback and to get more people involved it's something it's not the most of authority for some people for the most interesting fields working but I can assure it's uh it's and also place and we all nice people conclusion about writing the death of stuff of which we experience that we should never do things we did to the impression that we want to force stuff coming from the known 1 can 1 can be your can for KDE on no uh itself natural but it's really easily done to get to come to talk in a way that people of the sensor people that they are no would get stuff which was developed the known and then is dumped KDE with argumentation well the standard please the woman that stuff why because it's the standard so now we have to we have to take everyone with you if you want to succeed with something of yeah on the contrary is sometimes you need someone to just go ahead with the project have the other's full that's himself was stalled for a very long time unfortunately you know because the there was not enough people working on its hands at there was some stuff missing we had some some issues with the canonical seal a for example to get the suffers from available and what is division which in the end failed so working on so post created but it was just and and we also needed fedora to really push for upstream in order to make it happen under more distributions also doesn't make sense to invest too much time and by extending about details of in the same manner so bishops just go at was implementing thought something and users were complain about that later and then you can adjust it so there is that of a debating about hypothetical stuff beforehand and so benefits and enough questions they are microphone appears who voted questions to the user so it ends up on the video the the will take questions and more rows shows you the years questions when you move this in and there's the least unrepentant to provide a boost in stone Greenstone hooks like beyond that which is from simple and it provides a the package itself is not able to write any strip because the executed with the repair mission and therefore we don't want this during installation phase that the the packages able to execute any script which is happening is that the solid will I was scanned what is available in the package and then decide what is the best thing to do for example if it ships that are file it was later updated the death of database so is doing that and if the of the person who created all solid package um wants to do something which he or she would need a script for of they could just follow back again so style and we will implement that for every for every other person as well in the solid directly thank you very much for the upstream it's really really needed to this and up to them break you we need some tools to protect us from distributions them and Danish selection for the source of with the opposite the half of them my question is about this a DP 1192 school like this new universe sold Debian thing so what were the school teachers does more about this process is ongoing and confidence in our well it's a deadly things on and the appears that Odevene extension project and there was the this file will include information about components of the component data I talked about on the solid side but also information about the applications and stuff like Codex or a channel modules so we can if something is missing a fit of the cation misses a module which is would be necessary to work for example if you have a video converter which is to convert something but some gestural modulus missing to do that it can do make a packet get query on how to find out what which package provides the missing component again afterwards install it and so what you do with that level is to push all this metadata into 1 file and make it fair make it available on every distribution and evidently distribution using apps and enabling this feature which you it is just for deviant the 5 form is is is independent of course everyone can users from but the main thing is mainly about deviance and currently working on implementing that stuff in the devil lots of kids so I hope this will get ready soon we have 1 more question we have time for 1 more
yeah locally greatly during much if
I no area I think it was way too
fast but I was going to ask you you know you're tired at the time of the people so it will be very close to the talks you've got in the school community in relation the people of the talk will be doing presentation should actually like this in the top of the program here in and so you
inclusion not what the
of of the of the of stimuli but the book
App <Programm>
Stereometrie
Parametersystem
Streaming <Kommunikationstechnik>
Vorlesung/Konferenz
Projektive Ebene
Softwareentwickler
Frequenz
Computeranimation
Distributionstheorie
Bit
Prozess <Physik>
Extrempunkt
Datensichtgerät
Atomarität <Informatik>
Adressraum
Gruppenkeim
Kartesische Koordinaten
Übergang
Streaming <Kommunikationstechnik>
Metadaten
Deskriptive Statistik
Client
Datenmanagement
Existenzsatz
Translation <Mathematik>
Kontrollstruktur
Skript <Programm>
Vorlesung/Konferenz
Wurzel <Mathematik>
Phasenumwandlung
Umwandlungsenthalpie
Nichtlinearer Operator
App <Programm>
Addition
Statistik
Dokumentenserver
Kategorie <Mathematik>
Datenhaltung
Güte der Anpassung
Stellenring
Quellcode
Bitrate
Bildschirmsymbol
Elektronische Unterschrift
Teilbarkeit
Konfiguration <Informatik>
Teilmenge
Kollaboration <Informatik>
Dienst <Informatik>
Menge
Ganze Zahl
Server
Dateiformat
Projektive Ebene
Information
Ordnung <Mathematik>
Rückkopplung
Gruppenoperation
Interaktives Fernsehen
Sprachsynthese
Punktspektrum
Multiplikation
Bildschirmmaske
Software
Datentyp
Programmbibliothek
Elektronischer Fingerabdruck
Luenberger-Beobachter
Installation <Informatik>
Zusammenhängender Graph
Inhalt <Mathematik>
Maßerweiterung
Datenstruktur
Speicher <Informatik>
Softwareentwickler
Bildgebendes Verfahren
Widerspruchsfreiheit
Leistung <Physik>
Schätzwert
Benutzeroberfläche
Eindeutigkeit
Physikalisches System
Elektronische Publikation
Partitionsfunktion
Office-Paket
Offene Menge
Stereometrie
Distributionstheorie
Vektorpotenzial
Maschinensprache
Deskriptive Statistik
Metadaten
Client
Dämpfung
Trennschärfe <Statistik>
Kontrollstruktur
Skript <Programm>
Vorlesung/Konferenz
Schnittstelle
Metropolitan area network
App <Programm>
Computersicherheit
Güte der Anpassung
Napster
Biprodukt
Software Development Kit
Entscheidungstheorie
Generator <Informatik>
Dienst <Informatik>
Menge
Forcing
Festspeicher
Grundsätze ordnungsmäßiger Datenverarbeitung
Ordnung <Mathematik>
Subtraktion
Stabilitätstheorie <Logik>
Mathematisierung
Domain-Name
Programmbibliothek
Installation <Informatik>
Datenstruktur
Inklusion <Mathematik>
Maßerweiterung
Konfigurationsraum
Binder <Informatik>
Elektronische Publikation
Debugging
Wort <Informatik>
Dämon <Informatik>
Resultante
Bit
Prozess <Physik>
Punkt
Atomarität <Informatik>
Natürliche Zahl
Selbstrepräsentation
Versionsverwaltung
Kartesische Koordinaten
Element <Mathematik>
Binärcode
Übergang
Eins
Datenmanagement
Prozess <Informatik>
Bildschirmfenster
Umwandlungsenthalpie
Konstruktor <Informatik>
Schnelltaste
Freier Parameter
Dokumentenserver
Datenhaltung
Abstraktionsebene
Quellcode
Elektronische Unterschrift
Teilbarkeit
Konfiguration <Informatik>
Teilmenge
Texteditor
Projektive Ebene
Identifizierbarkeit
Information
Verzeichnisdienst
Schlüsselverwaltung
Telekommunikation
Gruppenoperation
Zellularer Automat
Zahlenbereich
Implementierung
Sprachsynthese
Kombinatorische Gruppentheorie
Physikalische Theorie
Code
Kappa-Koeffizient
Textur-Mapping
Zusammenhängender Graph
Softwareentwickler
Drei
Mobiles Endgerät
Programmiersprache
Einfach zusammenhängender Raum
Beobachtungsstudie
Kreisfläche
Benutzeroberfläche
Validität
Systemverwaltung
Physikalisches System
Quick-Sort
Künstliches Leben
Office-Paket
Integral
Objekt <Kategorie>
Mapping <Computergraphik>
Modallogik
Flächeninhalt
Mereologie
Hypermedia
Axiom
Verkehrsinformation
Distributionstheorie
Stereometrie
Prozess <Physik>
Versionsverwaltung
Kartesische Koordinaten
Binärcode
Videokonferenz
Übergang
Metadaten
Datenmanagement
Trennschärfe <Statistik>
Kontrollstruktur
Skript <Programm>
Vorlesung/Konferenz
Umkehrung <Mathematik>
Default
Phasenumwandlung
Softwaretest
Umwandlungsenthalpie
Parametersystem
App <Programm>
Dokumentenserver
Datenhaltung
Abfrage
Quellcode
Entscheidungstheorie
Datenfeld
Transcodierung
Projektive Ebene
Information
Ordnung <Mathematik>
Standardabweichung
Fitnessfunktion
Rückkopplung
Stabilitätstheorie <Logik>
Subtraktion
Division
Task
Datensatz
Multiplikation
Bildschirmmaske
Informationsmodellierung
Bereichsschätzung
Zusammenhängender Graph
Inklusion <Mathematik>
Softwareentwickler
Maßerweiterung
Grundraum
Autorisierung
Diskretes System
Physikalisches System
Elektronische Publikation
Modul
Energiedichte
Basisvektor
Flächeninhalt
Relativitätstheorie
Vorlesung/Konferenz
Optimierung
Kombinatorische Gruppentheorie
Vorlesung/Konferenz
Inklusion <Mathematik>

Metadaten

Formale Metadaten

Titel AppStream & Listaller
Untertitel The next step in application management and deployment on Linux
Serientitel FOSDEM 2014
Autor Klumpp, Matthias
Lizenz CC-Namensnennung 2.0 Belgien:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/32487
Herausgeber FOSDEM VZW
Erscheinungsjahr 2014
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract AppStream provides a solution for application-centric software management using existing package managers, while Listaller extends the package manager with the ability to install 3rd-party applications in a secure way, without introducing additional UI. This talk explains the basic concepts of both projects and the motivation behind them, as well as the obstacles in cross-distro collaboration which we hit while developing these tools. Software management on Linux is traditionally done using packages, although desktop users are more interested in applications instead and do not care much about how something is packaged. Also, users often want to install software which is not available in the repositories (e.g. unpackaged stuff, as well as new versions of applications), without having to upgrade the whole distribution. AppStream provides a solution for application-centric software management using the existing package-managers, by providing extra metadata for packages. This allows developers to create cross-distro application-centers, like GNOME-Software or Apper. Listaller extends the package manager with the ability to install 3rd-party applications, which are not (yet) in the distribution's repositories. It builds on top of the AppStream specs and focuses on integration, so the installation process is completely transparent for users, and Listaller-installed applications can be managed with existing software-management tools. Because software from 3rd-party sources is a potential security risk, Listaller also tries to give users a hint if they can trust a certain application, and includes sandboxing for these applications. Both projects rely heavily on metadata provided by upstream projects and distributions. This talk will explain the concepts of AppStream and Listaller, which new metadata needs to be added and why as well as how distributions can make a software-center available for their users. If the time allows it, the talk will also highlight the problems we had with the AppStream project in the first place and how we could improve cross-distribution collaboration in general

Ähnliche Filme

Loading...
Feedback