Announcement

Collapse
No announcement yet.

What Are The Biggest Problems With Linux?

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

  • What Are The Biggest Problems With Linux?

    Phoronix: What Are The Biggest Problems With Linux?

    If you were asked what are the biggest problems with Linux, what would that be?..

    http://www.phoronix.com/vr.php?view=MTExNzQ

  • #2
    This: Why Linux is not (yet) Ready for the Desktop (a.k.a. Linux problems), 2012 edition

    There's one problem no amount of money can solve: for most Open Source developers Linux is a playground, a thing they don't care beyond their aspirations, thus we have constantly broken features and API breakage every odd moon cycle. With such an attitude there's no way Linux will ever attract a big number of serious ISVs. Of course, people will be quick to point that already available Open Source software is enough for everyone - but that's a serious myopia. No, it's not enough, very very far from that.
    Last edited by birdie; 06-10-2012, 11:10 AM.

    Comment


    • #3
      Reduced complexity of the various distributions and software stacks.

      I don't understand why for example archlinux already installs three different ciphering libraries, libssl, libgcrypt and another one I forgot if you don't even install a user interface. Of course they don't exactly do the same, but it should be possible to reduce code which was implemented two times (f.e. the des algorithms) and install only one library...

      Another thing I always wonder about is why every linux distribution installs libgtk2.32 even so the user interface is the gnome-shell which is based on libgtk 3.0. It can't be that difficult to port the few tiny tools to gtk3? (isn't libreoffice now ported to gtk3?)

      And the last thing I won't to mention is that it doesn't make sense that every linux distribution uses their own thousands of scripts. F.e. in the case of the start-up scripts the distributions could use systemd and write good scripts only once and use the same ones in ubuntu/lfs/gentoo/arch...? In the end they all do exactly the same job and this would reduce a lot of work.

      Just my 2 cents...

      nice regards
      Michael

      Comment


      • #4
        "Linux" as a kernel, is very good I think. Central concepts such as multitasking and scheduling are working very well now. Low latency operation really sets the feel of the entire operating system, and I do believe a modern OS, can be combined with the feel of vintage assemblycoded hardware.

        Distibution-wise it needs to become even more professional. Less obscurity, more professionality. Meaning, no strange error msgs, no unesscesary obfuscation, no strange rules etc. You should boot it up, and expect a well-working system for professionals, without geekery.

        As an audio-professional, I wouldn`t mind a software package like Logic Audio either. I have already made a few plugins. (https://sourceforge.net/projects/pxu/files/) (audioexample, -5dB master)

        I would love to do my work on linux.

        Other than that, Doom 3 is playing like a vintage asm-coded arcade here. Gnome 2 was also working very well to my tastes. Could be that these people come up with something even better ofcourse. Else LXDE, or maybe Enlightenment might be good alternatives.

        If Ubuntu tightens up their act somewhat, maybe Mark stops discussing "lust" on his webpage, and related corporate philosophy, it could be good.

        Peace Be With You!
        Last edited by Paradox Uncreated; 06-10-2012, 11:20 AM.

        Comment


        • #5
          In my opinion two issues:

          1. Better + more complete (meaning source and usage) driver documentation.

          2. Mandatory, elaborate metainformation for binaries (like a .desktop file but also including e.g. dependencies)

          Comment


          • #6
            Originally posted by aelo View Post
            Another thing I always wonder about is why every linux distribution installs libgtk2.32 even so the user interface is the gnome-shell which is based on libgtk 3.0. It can't be that difficult to port the few tiny tools to gtk3? (isn't libreoffice now ported to gtk3?)l
            You mean such tiny tools like GIMP?

            And the last thing I won't to mention is that it doesn't make sense that every linux distribution uses their own thousands of scripts. F.e. in the case of the start-up scripts the distributions could use systemd and write good scripts only once and use the same ones in ubuntu/lfs/gentoo/arch...? In the end they all do exactly the same job and this would reduce a lot of work.
            Isn't that why there are different distributions in the first place? Because not every one wants to use the same stuff? I hope that things like systemd or pulse-audio never are standard for my distribution, but this may be different for other people.

            For me the biggest problem of Linux is being a second class citizen for almost any hardware manufacturer when using Linux.

            Comment


            • #7
              The biggest problem with Linux is probably the API and ABI changing almost every day and new features and software expecting that new API/ABI. MS invests a lot of time and money on maintaining a stable API and ABI , only breaking it on rare occasions. I'm a maintainer for a small Linux distribution, Frugalware and I see the problems we are face with every day ... You need to use a new version of libX but using a new version of libX will break progY while progZ still needs it. Also the new version of libX will expose a bug in progZ and so on...
              Red Hat tried to do something with RHEL to keep things stable in the long term but ... you can install a driver for the latest Radeon GPU and have it work with Windows XP from 10 years ago. You need the latest kernel and even then it won't probably work right in Linux.
              With decent HW and SW support I think Linux could start to gain some market share but someone needs to start it all and invest and talk to the open source community , collaborate with them. What I have seen working are things like WebKit. It was open source at first , then it was embraced by Apple and now it's the basis of the world'd most successful browser ( Chrome ). For open source and Linux to really succeed it's likely that a company has to take it and develop it in-house while at the same time collaborating with interested open source developers, companies. Google did so with Android and it's a success no matter what people say, let's not forget that they did have to fork the kernel ...

              Comment


              • #8
                Two long running, common issues

                There are two common issues that have been hanging around for a long time:

                1. Nouveau doesn't sync to vblank when using a compositing window manager, causing tearing.
                2. The Realtek r8169 driver bounces the connection periodically.

                This is very common hardware. The workaround for both is to install the vendor driver, which is fine for most of us, but not end-users.

                I understand the limitations that the developers are facing, and recognize that they are working to get functionality as quickly as possible. However, there's tendency to concentrate on issues which, while important, are not user visible. No matter how great the internals are, if the basic visible functionality stays broken for years, end-users do not perceive improvement.

                Comment


                • #9
                  Originally posted by birdie View Post
                  This: Why Linux is not (yet) Ready for the Desktop (a.k.a. Linux problems), 2012 edition

                  There's one problem no amount of money can solve: for most Open Source developers Linux is a playground, a thing they don't care beyond their aspirations, thus we have constantly broken features and API breakage every odd moon cycle. With such an attitude there's no way Linux will ever attract a big number of serious ISVs. Of course, people will be quick to point that already available Open Source software is enough for everyone - but that's a serious myopia. No, it's not enough, very very far from that.
                  most of the points in this article are because gnu/linux has a small market share :-P (no good consumer gfx-driver support, adobe flash, ...)
                  and why is it gnu/linux fault that there are patented technologies they cant implement?
                  also fontrendering? elementary os for example has much nicer look of the fonts than windows :-P

                  yes there are some problems, but they are either being worked on (wayland as replacement for xorg) or are not the fault of the developers (microsoft doesnt release a kick ass photo editing programm either but on foss-plattforms have to? :-P)

                  Comment


                  • #10
                    the desktops suck. kde is an unholy mess that gets in the way. gnome is too watered down. unity is like a fugly version of gnome.
                    the enterprise distros like rhel ship with broken opensource video drivers. this gives the impression that linux sucks at video, when the proprietary drivers work fine. this is not fit for an "enterprise" OS, where moving a window may cause a kernel panic. Only ship stable software. That means use vesa, until nuveau is fixed.
                    the GUI managment tools are very lacking. this basically caused Novell to fail at their plan to evangelize linux into offices. no "non-geeky" sysadmin is gonna want to edit goshdarned text files in vi every time he has to do some administration.
                    I think linux in offices in the next area where Red Hat and Attachmate needs to really focus. That means making OpenOffice something which does not cause me do driver a fork in my eye.
                    http://tmrepository.com/trademarks/y...helastversion/
                    http://tmrepository.com/trademarks/b...ceddeathmarch/

                    Comment


                    • #11
                      DansGuardian needs some work and a front-end for management. Content filtering is required in many schools and libraries and such filters are readily available on Windows. I wrote some utilities with bash scripts and dialoggers but something more polished is needed. There's a Webmin module but it's only slightly better than a text editor.

                      There are some GUI front-ends available but they have many limitations like a lack of support for multiple filter groups, specific proxy requirements (like tinyproxy), and no easy way to enable proxy anti-bypass firewall rules and network bulletin boards (informing users of what they are doing wrong when attempting to bypass the proxy).
                      Last edited by jhansonxi; 06-10-2012, 11:43 AM.

                      Comment


                      • #12
                        Originally posted by EvilTwin View Post
                        most of the points in this article are because gnu/linux has a small market share :-P (no good consumer gfx-driver support, adobe flash, ...)
                        and why is it gnu/linux fault that there are patented technologies they cant implement?
                        also fontrendering? elementary os for example has much nicer look of the fonts than windows :-P

                        yes there are some problems, but they are either being worked on (wayland as replacement for xorg) or are not the fault of the developers (microsoft doesnt release a kick ass photo editing programm either but on foss-plattforms have to? :-P)
                        The mentioned problems are getting fixed all the time the problem is that I see no end in sight. Besides, if something doesn't work in Linux, people won't f*cking care whose fault it is, "It works in Windows/MacOS/whatever - Linux sucks", and they are right.

                        Comment


                        • #13
                          Problems with configuration

                          Without a doubt I think the number one problem for many things Linux is one of configuration and software/hardware -- for anything from PostgreSQL to MySQL to SSD tuning mdadm formatting, the defaults are not exactly ideal, the options are many, and finding the right ones is a pain.

                          When you install a distro on an SSD, it should know to use trim (discard), and maybe your distro should have considered whether it's better to try default, commit=10, or even commit=600 (in some debian docs). When you create a disk with mdadm should the user really need to figure out the minutia of parity alignments or realize that they need to manually set the chunk size to something sane? With MySQL and PostgreSQL there are slews of cache and memory and performance settings you might need to tweak.

                          Name a large or low level piece of Linux and the distro defaults probably fall into the category of 'will work, but kind of stupid.'

                          Of course there are some things that fall into the category of hassle... For example, ever try to monitor your temps under linux?
                          You'll probably need at least 3 commands: something like aticonfig --odgt, sensors, and to either have set up or invoked several time hddtemp. Even worse, the output from sensors will likely be mostly wrong unless you dig through it and figure out oh, what it reports as fan one is really my PWM case fan, and fan 3 is my CPU fan2, and none of their number match what's in bios, and in some cases you'll even get numbers that seem non-sensible. And good luck figuring out which temp goes to what.

                          Now none of these things are insurmountable. They're just stupid hassles that the end user ought to be able to not have to figure out.

                          I'm barely scratching the surface here, but the gist of it is that default configurations need to be contextual and way smarter. It might even make sense to, horror or horrors, have a wizard or something now and again.

                          Also, just a shoutout to that section of linked articles that mentions the sound system... I haven't been able to figure out what the deal is exactly, but what which audio port is which sure doesn't seem to match the mobo manual. Or something. In general, though, even when it's working correctly, there's a huge variation in volume normalization and output between programs. If I leave system volume at 50%, I'll often have to turn my speakers down to play music and up to watch a movie. No idea why.
                          Last edited by sloggerKhan; 06-10-2012, 12:15 PM.

                          Comment


                          • #14
                            Originally posted by garegin View Post
                            the desktops suck. kde is an unholy mess that gets in the way. gnome is too watered down. unity is like a fugly version of gnome.
                            the enterprise distros like rhel ship with broken opensource video drivers. this gives the impression that linux sucks at video, when the proprietary drivers work fine. this is not fit for an "enterprise" OS, where moving a window may cause a kernel panic. Only ship stable software. That means use vesa, until nuveau is fixed.
                            the GUI managment tools are very lacking. this basically caused Novell to fail at their plan to evangelize linux into offices. no "non-geeky" sysadmin is gonna want to edit goshdarned text files in vi every time he has to do some administration.
                            I think linux in offices in the next area where Red Hat and Attachmate needs to really focus. That means making OpenOffice something which does not cause me do driver a fork in my eye.
                            http://tmrepository.com/trademarks/y...helastversion/
                            http://tmrepository.com/trademarks/b...ceddeathmarch/
                            From the sounds of it, I don't suppose you like the Windows or Mac desktops either. And sysadmin under Windows is the biggest f***ing pain I've ever come across extremely difficult automating stuff and every single driver/piece of softwhere you instell gets in front of your face asking something or other.

                            No, what linux needs is an easy way for developers to reliably release their apps to linux distros. If for example you look at the humble bundle games, many of them need a bit of tweaking to get sound to work. I've no idea how well Ubuntu software centre works, but if it's Ubuntu-only, it's not enough. Linux is fragmented into many distributions because distribution is what linux does best. That, and a constantly moving target, make "just making things work" harder than with MS or Mac, but not impossible!

                            Comment


                            • #15
                              Caving to the lowest common denominator, following bullshit ui/market trends.

                              Comment

                              Working...
                              X