Announcement

Collapse
No announcement yet.

Flatpak Officially Announced For "Next Generation Linux Applications"

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    Originally posted by Mystro256 View Post
    Well, two major issues I see with snap: less secure and bigger file sizes. As for Flatpack, I'm pretty sure the only downside is that it requires systemd (for now).
    Except the systemd requirement isn't a downside.

    Comment


    • #42
      Originally posted by debianxfce View Post
      When you google "4G network hacked", you do not find any or many sagas from 2016. Of course security holes are fixed in 4G networks like in any advanced computing system. In high tech countries 4G has replaced adsl, so a lot of desktops are using 4G for example via 4G router that has a firewall.
      I'm still not getting how 4G is better than ADSL or fiber. Modem/routers have the same firewalls as decent 4G routers. It sucks balls as in most embedded devices.
      And 4G can be hacked remotely while ethernet hacking is a bit more involved.
      Hacking the infrastructure is the same for both.

      Comment


      • #43
        Originally posted by johnc View Post
        Umm... deciding which LIBRARY you use in your application is completely different than having two sets of code for each library so you can target multiple distributions, which is what you said. Kodi never used libav. It was always an ffmpeg project. Nor did they ever have a libav version of Kodi and an ffmpeg version. Kodi didn't even use system-installed libraries (nor did mplayer2), though XBMC did once make that option available. mplayer2, once they went to libav, was never again tied to ffmpeg, and libav was built into the package. The guy who maintained mplayer2 never had anything to do with ffmpeg once it was converted over to libav. Kodi and mplayer2 are cross-platform projects that bundle what they need right into their own packages. Exactly as Mez' said before you threatened to "punch him in the throat".

        You seem to be conflating "having to choose between forks of a similar library" to being the same thing as needing these "flatpak" packages to target multiple Linux distributions.
        Then why the hell I cannot just go on Kodi site and pull down a tar-gz of things, unzip that in /opt and run it from there?

        Kodi uses system libraries, try to install it in Suse without pulling and replacing system libraries with those from packman third party repo (hint: it segfaults), and same on Debian, it needs system libraries, cannot live without.

        Comment


        • #44
          Originally posted by debianxfce View Post
          WAN IP changes after every restart of 4G client, no fixed IP->no attackers.
          this happens with at most daily frequency on most cable connections too. I have to pay a bonus to get a fixed IP from any known ISP.

          I'm not getting why changing IP increases security of clients anyway.
          The attacks on clients are mostly through browser or applications and install rootkits/trojans that will then advertise the node whatever IP it has.

          DDOS is a denial of service, it's done on servers, not clients.

          Comment


          • #45
            Originally posted by computerquip View Post

            This makes me want to punch you in the goddamn throat for not even having an inkling of what you're talking about while stating things as fact.

            1. This isn't the windows model. If any comparison can be made, it's to the MacOSX bundle in a sandbox.
            2. You've clearly never developed or deployed an application in your life. If you deploy on Ubuntu and it doesn't support a library, what do you do? This is the *entire* reason why the libav and ffmpeg debacle was such a ridiculous mess. Applications can't just bundle the library they want to support. They instead had to have two branches of code to support both or simply not support some distributions that didn't support one or the other.
            3. If you make the argument of, "Well the user can just go through a third party to get the libraries", name one other platform that does this. You wouldn't (and shouldn't) even do this on Windows. How do you know that the library wasn't configured in such a way for that application? How do you know the version that you have is secure?
            4. As it is now, even if a library is supported by the distribution you're supporting, it may not be compiled in a way that you want it to be to support a certain feature or such.
            5. Windows and MacOSX isn't safe when executing third party applications, in that each application can potentially do damage to your system. Linux applications are the same way, even though they're made a little bit more trustworthy by going through a signed package. Flatpak is the most ideal situation where the application has access to only what it needs and nothing more. It can't do any arbitrary damage.

            It's absolutely ridiculous how some people don't see the purpose of flatpak. Have you been living under a rock? Or perhaps you'd like to make sure that Linux as a desktop has the qualities of a rock?
            Not gonna answer this since the tone used is really agressive and condescending.

            Originally posted by computerquip View Post
            Also, lazy half-involved parties such as Valve? They're the reason we have any major push for gaming on Linux. They're a good portion of the reason we have Vulkan right now and a lot of LunarG's work was funded directly by Valve. How can you possibly call them "lazy half-involved"? What have you contributed in comparison?
            Here, I'd like to be more specific about the reason I said this.
            I'm not talking about them bringing us games (I'm glad about that obviously), if you weren't flying off the handle, you would understand I was mainly talking about the libraries, and there is indeed a very real problem of libraries (namely libstdc, libgcc and libxcb) within Steam that actually breaks it. Meaning workarounds to be able to launch steam with open drivers. After 3 years, this should have been tackled to a cleaner solution.

            This is exactly the kind of issues pending with flatpacks/snaps.

            Comment


            • #46
              Originally posted by debianxfce View Post
              Here we had fixed wan ip for free with adsl.
              Developing country? AFAIK only developing countries have that for free, and good 4G contracts.
              This makes me think they also didn't filter the line in any way.

              Use changing wan ip and monitor you firewall logs->no attackers.
              Until the IP/port scanners reach your new IP anyway. Turn off anything that might attract them, ping, ssh, telnet, whatever.

              For 3G/4G all modem/routers I see around lack any decent feature nor are kept updated/safe I use a router with usb port and openwrt/LEDE firmware and a dumb 4G usb stick controlled by it.

              That is windows users feature, not Debian users.
              This is how hacking clients works, linux is not a target usually, but a client is a client is a client. If you want to attack a client you don't DDOS, as that only locks up their connection.

              4G router is a server
              No, it would be called server then. If it is called router it's because it is not a server.

              4G operators do have resources to fix any attacks to their system, usually it is a free update from 4G infrastructure manufacturer..
              4G is network infrastructure, it only moves data around. It is not different from whatever else.

              I think the only difference is that your new ISP actually placed serious firewalls to stop the low-end IP scanners and low-end crap from wasting their bandwith, while the idiot ISP with cable didn't (or you didn't pay for their firewall services), and you got the same IP/port scanner crap that hits all unfirewalled public addresses.

              Comment


              • #47
                does it rub anyone else the wrong way that flatpak stores architecture dependent data files in ~/.local/share? if ~/.local/share was/is intended to be a user specific version of /usr/share, then this is a flagrant violation of the linux fhs. if it wasn't, then why would the ppl who came up with the xdg base dir standard use the name ~/.local/share in the first place, rather than something like ~/.data or ~/.local/data instead? imo either way you look at it it doesn't make much sense

                Comment


                • #48
                  Originally posted by debianxfce View Post
                  In our router, we had no attackers with 3G with different operator, so 3G and 4G networks are safer than adsl that we used 4 year ago from 2 different ISP.
                  As I said, 3G and 4G network infrastructure have much more limitations (not just bandwith), so the ISP MUST place decent firewalling before stuff goes in their infrastructure.

                  A DDOS on a device on 4G infrastructure would overload a whole cell and block the 4G connection of half a city, and that's not acceptable.

                  So it's not "3G and 4G" per-se, it's just that the ISP is firewalling for you, and for these networks the ISP MUST firewall.

                  Cable internet can be firewalled from ISP too, as I said. In your case, 4G is better, but it is NOT universally better.

                  Comment


                  • #49
                    Originally posted by Mystro256 View Post
                    Well, two major issues I see with snap: less secure and bigger file sizes. As for Flatpack, I'm pretty sure the only downside is that it requires systemd (for now).
                    i dont see why the snap approach is less secure than the flatpak aproach.
                    if you refer to the 1GB libreoffice snap package: https://skyfromme.wordpress.com/2016...reoffice-snap/

                    Comment


                    • #50
                      Originally posted by e_tank View Post
                      does it rub anyone else the wrong way that flatpak stores architecture dependent data files in ~/.local/share? if ~/.local/share was/is intended to be a user specific version of /usr/share, then this is a flagrant violation of the linux fhs. if it wasn't, then why would the ppl who came up with the xdg base dir standard use the name ~/.local/share in the first place, rather than something like ~/.data or ~/.local/data instead? imo either way you look at it it doesn't make much sense
                      AFAIK flatpack can install stuff for single users or for everyone, if you install for a specific user it makes sense to store files in there. Where does it install things in if you install for all?

                      Comment

                      Working...
                      X