Has modern Linux lost its way? (Some thoughts on jessie)

For years, I used to run Debian sid (unstable) on all my personal machines. Laptops, workstations, sometimes even my personal servers years ago ran sid. Sid was, as its name implies, unstable. Sometimes things broke. But it wasn’t a big deal, because I could always get in there and fix it fairly quickly, whatever it was. It was the price I paid for the latest and greatest.

For the last number of months, I’ve dealt with a small but annoying issue in jessie: None of Nautilus, Thunar, or digikam (yes, that represents Gnome, XFCE, and KDE) can mount USB drives I plug in anymore. I just get “Not authorized to perform operation.” I can, of course, still mount -o uid=1000 /dev/sdc1 /mnt, but I miss the convenience of doing it this way.

One jessie system I switched to systemd specifically to get around this problem. It worked, but I don’t know why. I haven’t had the time to switch my workstation, and frankly I am concerned about it.

Here’s the crux of the issue: I don’t even know where to start looking. I’ve googled this issue, and found all sorts of answers pointing to polkit, or dbus, or systemd-shim, or cgmanager, or lightdm, or XFCE, or… I found a bug report of this exact problem — Debian #760281, but it’s marked fixed, and nobody replied to my comment that I’m still seeing it.

Nowhere is it documented that a Digikam mounting issue should have me looking in polkit, let alone cgmanager. And even once I find those packages, their documentation suffers from Bad Unix Documentation Disease: talking about the nitty-gritty weeds view of something, without bothering to put it in context. Here is the mystifying heading for the cgmanager(8) manpage:

cgmanager is a daemon to manage cgroups. Programs and users can make D-Bus requests to administer cgroups over which they have privilege. To ensure that users may not exceed their privilege in manipulating cgroups, the cgroup manager accepts regular D-Bus requests only from tasks within its own process-id and user namespaces. For tasks in private namespaces (such as containers), SCM-enhanced D-Bus calls are available. Using these manually is not recommended. Rather, each container is advised to run a cgproxy, which will forward plain D-Bus requests as SCM-enhanced D-Bus requests to the host cgmanager.

That’s about as comprehensible as Vogon poetry to me. How is cgmanager started? What does “SCM-enhanced” mean? And I even know what a cgroup is.

This has been going on for months, which has me also wondering: is it only me? (Google certainly suggests it’s not, and there are plenty of hits for this exact problem with many distros, and some truly terrible advice out there to boot.) And if not, why is something so basic and obvious festering for so long? Have we built something that’s too complex to understand and debug?

This is, in my mind, orthogonal to the systemd question. I used to be able to say Linux was clean, logical, well put-together, and organized. I can’t really say this anymore. Users and groups are not really determinitive for permissions, now that we have things like polkit running around. (Yes, by the way, I am a member of plugdev.) Error messages are unhelpful (WHY was I not authorized?) and logs are nowhere to be found. Traditionally, one could twiddle who could mount devices via /etc/fstab lines and perhaps some sudo rules. Granted, you had to know where to look, but when you did, it was simple; only two pieces to fit together. I’ve even spent time figuring out where to look and STILL have no idea what to do.

systemd may help with some of this, and may hurt with some of it; but I see the problem more of an attitude of desktop environments to add features fast without really thinking of the implications. There is something to be said for slower progress if the result is higher quality.

Then as I was writing this, of course, my laptop started insisting that it needed the root password to suspend when I close the lid. And it’s running systemd. There’s another quagmire…

Update: Part 2 with some reactions to this and further thoughts is now available.

124 thoughts on “Has modern Linux lost its way? (Some thoughts on jessie)

  1. This is a security risk. If people get so comfortable with having to give their password all the time they will eventially just do it without thinking. We know it’s never a good idea for people to start typing passwords without stoping to think “Who wants my password and why?” So why would we encourage this? Also the more times a password is entered the easier it is to be captured by malware, and such.


  2. Thank you John for this opinion, I can totally relate.

    A cause for this problem is adding layers on top of layers and hacking the lower layers, instead of implementing the necessary leaks in the upper abstraction layers and the features to support them in the lower layers. Note that this would require communicating between projects *before* hacking.

    I was trying to change the primary DNS for all connections on my workstation yesterday, didn’t even know where to start: resolv.conf? Overwritten by dhclient. dhclient config scripts? Overwritten by network-manager. Network-manager then? Doesn’t have an option to prepend a DNS before those given by DHCP. Interestingly, many people already wrote scripts to hack network-manager which already hacks on top of dhclient which hacks on top of resolv.conf. Crazy and unsustainable.


  3. I moved to FreeBSD for several reasons. Native and well integrated ZFS support was one additional major factor. Having a simple and stable self-contained base system was another. The Debian base is now essentially out of our hands–we’re tied into doing what systemd does, with little other options. This is a worry given its continued scope creep, even if we’re happy with it today (though I’m personally not happy). I thought it would be a pain but modern pkg(8) is getting on a par with apt-get including a SAT-solver. While using a desktop environment might pull in dbus and the rest of the mess, at least it’s optional for the most part. None of the junk is present on my servers, and they remain simple and understandable, and likely more reliable.

    At the system level, I’ve been increasingly frustrated by the poor quality of all the new stuff. Even udev. Take something routine like making and removing LVM snapshots. It’s fraught with race conditions and can lock up your system. This isn’t due to LVM for the most part, it’s due to udev, and its helpers helpfully probing the newly-created block device. But this races badly with removal and other changes on a busy system. Result: it’s totally unreliable in the absence of safe coordinated locking. We had to abandon using LVM snapshots on the buildds for this reason. I don’t have this issue with ZFS snapshots. With all the increased complexity of everything connected by dbus and firing off events all over the place, this is likely to only become worse. I do worry about the security of all this stuff since it’s impossible to prove how all the authorisation layered on top of the regular stuff behaves. And the “session management” provided by logind/systemd is just a bad joke. I do wonder how much of the glue is really necessary, and how much is a workaround for not using ACLs; with ZFS on FreeBSD it has full NFS4 ACL support which makes a lot of things possible that simply can’t be done with Linux even when using (POSIX 1003.1e WITHDRAWN DRAFT) ACLs. Obviously not a solution for all the problems, but I can’t help but think that the current generation of solutions are a giant hairball of hacks, rather than a well considered and well implemented one.


  4. I do not know anything about computers I hate them. Now we have that out of the way.

    Linux and the Linux desktops are total nuisance! Samba works one minute and then suddenly stops working and then suddenly starts working. The Debian latest desktops are a total mess. I have had cleaner more workable desktops in the early 90s even the Amiga Workbench is cleaner. KDE misty Windows and special effects they hurt my eyes. I turn all that nonsense off because I know they are damaging my eyesight. The gnome desktop what has happened it is childish. Somebody logs onto the HP server and wants to print a document why do they need a root password! I spend hours making it work and when I get it to work just a little bit the way I would like it. They break it with a update, bug fixes security updates blah blah blah. I lose my network drives it makes me feel a little bit sick because I think I may have lost my data because they broke it with the update. I find my network drives are still there I just no longer have permission to access them. I spend ages setting them back just the way I like them. I squeeze the cheeks of my arse together because I think it may have deleted some data. I’m frightened to touch the fucking thing.

    SuSE has become openSuSE, and it’s broken it is not reliable in a work environment KDE as become a toy. Debian and gnome good grief what a pile of crap. Red Hat, overpriced monster and who is in control of my data me or them! or the NSA. Ubuntu is pathetic I could get more work done on a Amstrad PCW8512 word processor..

    USB stick unrecognised. Put it in a Windows system and it shows you what is on the USB stick and it shows you it has a hidden partition. Try again with Linux unable to mount unknown system. Start doing work on Windows systems instead. Purchase a virus scanner purchase a two-way firewall ransom-ware need to pay for them every year.

    Need to file a report multitasking copy and pasting as quick as possible before printing and posting. Gnome desktop suddenly turns into multiple icons because I have moved the mouse to the corner of the screen.. Oh my God where is my document gone! Squeeze the cheeks of my arse together my document is one of those little Windows thank goodness for that! Copy a large file over from the HP server Samba had a problem copying the file it deleted the file both ends because it attempted to copy the file to a new direction but fouled and deleted the original. Had a backup good grief wasn’t I lucky. I should not need to be lucky!

    Look to solve a problem on a search engine a load of antisocial idiots who have used copy and paste to copy somebody else’s answer which is opening a backdoor on a Linux system for a user outside user! They are all copying the same Samba, hack to their own system because they don’t know what they’re doing. Linux desktops have become shit. I could do more work on them in the 90s then I can do now. I now have 5 Linux systems and I don’t trust them. I now have a huge server running Windows with backup drives going everywhere. I don’t like Linux any more I don’t like using the computer with it on because I don’t know what it is going to break tomorrow.
    Thank you that was very satisfying to get that frustrated stress out.


  5. I’m a little confused. Are you saying that you’re surprised that Debian Unstable, is well, Unstable?


    Roger Leigh Reply:

    Franky, yes. My primary workstation has run Debian unstable since 2001; stable prior to that. So that’s ~14 years continuously tracking unstable and using it daily as a desktop system for development work. I reinstalled on two occasions when I switched from i386->powerpc and later to amd64, but I could have kept the same system for all that time. In all that time I saw two problems, both of which were transient and easily worked around.

    Note that Debian unstable doesn’t as a rule contain unstable unreleased software; that’s for experimental and often not even then. It’s only the packaging that’s possibly not quite right, and even then most developers took pains to ensure it was correct–after all it’s feeding into testing for the next release.

    So a change to its daily usability is indeed quite a shift. This isn’t “rawhide”.


  6. Well, you have to have a DE in FreeBSD also, right? Isn’t it going to be fundamentally the same as in Linux? You want KDE, you get the same KDE; you want Gnome, you get the same Gnome? How is the FreeBSD userland different from apt-get install xfce or something in Debian?


  7. I should have been more precise. I know FreeBSD has a different init system, a different libc, different tar, etc. But if we’re talking the desktop environment, isn’t it pretty much the same?


  8. @Pigsy – THANK YOU.

    To the rest of the UNIX|Linux (I don’t need to, nor care about pedantic crap of if being BSD, Debian, RedHat, Ubuntu, etc.) – If you are not thinking of the end-user – Joe & Jane Person who just wants to get their work done – you aren’t thinking!

    Back to computer fundamentals 098 (you should know this BEFORE you take fundamentals 101 your freshman year) – The role of a computer is to make your life easier. Not harder, more complex, more miserable. If you have some strange, S&M bent with your programming – take it somewhere else. From my perspective your creed isn’t built by how obfuscated and complex you’ve made it – but by how well you were able to take a complex issue and create a straightforward and intuitive solution for it. Yes, you can have the ‘for uber-geeks’ mode.. but for 80-90% of the use cases it should be pretty straightforward, clear, concise. With MEANINGFUL messages and documentation that doesn’t look like it came from a geeky grad student that failed all their English classes… starting with the 5th grade.

    Not a fanboy of Apple, but Steve J had the right idea. Consistency throughout the OS. Windows, at least through Win7 – same thing. Straight forward, fairly easy to use, AND IT doesn’t stop working mid-sentence.

    Hence – I no longer use any of the *nx’s. I deal with this crap all day at work… I don’t need to come home and deal with it there as well.

    Fred in IT


  9. There are 30-40 org.freedesktop-namespaced interfaces and services running on the average Linux desktop dbus. There are docs of varying quality available on freedesktop.org, but precious little that explains all their interactions and event dependencies. I wouldn’t be surprised if there are fewer than a couple of dozen people in the world who understand how they all hang together to make a desktop session. And that’s not even factoring in the desktop environments themselves! It’s a sad state of affairs.

    Worse, it all changes so often than any homework you do today to bring yourself up to date will probably be outdated in six months, maybe a year if you’re lucky.

    Mind you, people keep going on about systemd but it’s just another symptom of a longer-term problem that probably started around the time HAL appeared and disappeared. It can’t even be said this stuff is not necessary – it all is, even systemd. It’s just the way it gets iterated and shoved out into the world that leaves people feeling helpless from the ground constantly shifting beneath them.


  10. I was affected by the same issue with a fresh installed Ubuntu, the stable one. My only “sin” was to have started with a minimal install, and the installed XFCE. Nonetheless something that one expects to be working on a stable..but we know ubuntu has lost its soul to unity.

    I went through most of the same loops as yours, reading about polkit etc. in the end the fix was not to touch polkit or any of these new hateful packages but rather than to install some obscure udisks2-related package.

    Yes, just plain simple that. And I was tipped in this direction by a random guy that posted his solution on a forum I can’t remember where now…

    I am afraid it’s going down the toilet, and this is just the beginning..


  11. The last time I tried BSD it set the hwclock to localtime. This is nuts. It means BSD does not play well with other unices. There’s a long-time bugreport for this, no action.


    nikodemus Reply:

    That’s long history. Even the installer asks you wether your clock is UTC or local time.


  12. I’m watching GNU/Linux development for quite a time as an embedded engineer looking for a replacement of my RTOS by embedded GNU/Linux. I found that too many influence came from the desktop people, no matter if it was GNOME, KDE or others. The clean structure has gone and GNU/Linux is more like Windows than Unix now. The whole development is to desktop agnostic and there is also no such thing as as requirement or architecture specification for the developers. Systemd made things even worse. Thats the reson why I abandoned the idea of replacing RTOS by GNU/Linux. Instead I’ll use a multicore with simple static partitioning and run RTOS and GNU/Linux side by side.


  13. I’m experiencing this right now. I’m trying to build a pretty popular package and it’s giving me a generic error. Do I report a bug against the package? Against Bitbake? Is it a problem with Yocto? OpenEmbedded? Is it my configuration? Is it in one of the dependencies for the package? Is it in one of the dependency’s dependencies? Is it my firewall? Is it a problem at git? Is it a problem with the compiler. I haven’t got the foggiest idea *who* to report a bug to.


  14. Hey folks,
    I’ve been looking at all this stuff people have been grousing about Linux all across the Internet, and have concluded that a very large number of Linux users aren’t keeping up to date with a lot of the ‘new stuff’ coming up.
    Linux and Opensource are very quickly becoming ‘Big Business’ – with cloud and containers and so on. This has necessitated a need for much stricter privacy and access controls and better system admin tools to go with it.
    There are so many people apparently unwilling to put in the time to stay up to date – with things like systemd, policy kit, pam changes, avahi and mdns, FUSE updates, and so on …
    Linux is moving away from the ’90s and away from being a geek’s toybox. For me, its a good thing because I now have a much easier time getting clients to at least partially convert to Linux – servers and even desktops.
    Its similar to MS Windoze XP – people are still trying to stay in the past with it, and now there are lots of new things that don’t work with XP any more. People have to migrate to at least Windows 7 to remain viable.
    Suck it up guys and gals – there is a lot to learn and I agree-its not easy to keep up to date because people are too busy writing updates and new things, and leaving the documentation behind. However, this does not preclude the necessity to hunt things down – even as far as offering to help some of these projects get cleaned up so that everybody benefits.
    Its now an opensource world – please get with the program!


    Nate Bargmann Reply:

    Steve, with all due respect, the attitude you’ve chosen to use in your post is really what I see to be a major problem facing our community today. Telling us to “suck it up” is no different than the sort of arrogance I encountered with proprietary software well over 20 years ago. I did “suck it up”, I found Linux and left, or so I thought, that sort of attitude behind.

    I think that you will find that people like John have no aversion to learning new things. However, change for change’s sake strikes me more as a treadmill driven by marketing than a learning opportunity or born of necessity. Things seem more opaque now and too abstract plus the churn seems to be increasing with each passing year. How is this more understandable or secure?

    Speaking for myself, I do feel like as though I’m becoming less a master of and more of a slave to the computing software I use. This is not a good thing.


    Roger Leigh Reply:

    All very true. It seems like history repeats itself. Linux is now in the position proprietary Unix was 20 years back, and the arguments are just the same. I don’t like the attitude from Steve–it mirrors exactly what people were arguing *against* Linux when it was at the bottom of the pile.

    People have been arguing that *of course* only Linux could possibly be used professionally because alternatives would not be supported and come with commercial support. Go back in time and exactly the same thing was said about Linux, and yet people brought it into their companies and institutions and it superseded the supported expensive Unix systems because it filled a need (many needs), but in time became exactly what it replaced…. Some people, particularly corporate IT types, like this. Most of them are ignorant or willfully unaware of its history, many having come from Windows admin. But it’s a stark contrast from the goals and philosophy of the people who actually did the work to create it originally, and overall I think current development is too skewed in favour of corporate interests compared with the healthier balance we had 10-15 years back. I look at current Linux systems and I see sterile, sanitised corporateware.

    Steve’s comments above about keeping up is in exactly the same vein. Linux is becoming increasingly less diverse, and the notion that I must “keep up” implies that these “new” things are not only worth learning (most of the new stuff isn’t even documented, and will change in the next 12 months anyway) about but also that there are no alternatives to them which are also worth of using instead. Back when Linux was the plucky underdog, we went out of our way to make it the jack of all trades that did everything and was compatible with everything else. Filesystems: got the lot. File serving: interoperable with every system. Network protocols: Ditto. Now Linux is at the top of the heap, turning around and dictating that things must be done this way, and this way only, is to me a great shift in philosophy and not in touch with where we came from. It’s also quite hypocritical and dictatorial. We certainly didn’t appreciate it when it wasn’t us doing the dictating, so why should it be OK to do this now? It only breeds resentment and anger. That support for everything was what what made Linux useful to so many people–it could be the glue that tied everything together and serve all sorts of niche needs. Many of the new changes, including much of the userspace desktop glue and some parts of systemd, break compatibility with other systems. I’ve even seen increasing numbers of people arguing that POSIX no longer matters, and that Linux can go its own way and to hell with the rest. These people are idiots who are greatly overstating their own importance (and indeed of Linux in general). They remove what was the real meaningful difference between Linux and commercial Unix, at which point I’d have to look at whether it was worth sticking around at all if ultimately, this was all that became of the great hopes we had back in the 90s.


    Patrick Reply:


    would you agree, that this is a big chance of the BSD’s to get a bigger Community? I used FreeBSD 10 years ago and tried it again… and i see a lot of progress in their Operating System.

    Roger Leigh Reply:

    I think the BSDs will see a modest increase in the short term to be sure. I think with pkg-ng and freebsd-update it’s massively easier to install and upgrade than previously, and newer derivatives like PC-BSD do give you a desktop with almost as much polish as on Linux. That said, I’m sure that whatever happens, most users will stick with Linux in the short term; inertia alone would ensure that. That said, the real deciding factor is ultimately down to where the developers are. That will affect things in the longer term.

    John Goerzen Reply:

    Nate, that is really insightful, especially becoming a slave rather than a master. Good thoughts there.


    Steve Dupuis Reply:

    I guess people are missing the point – Linux is changing – big time. Its rise in popularity is because of its interoperability and depth and breadth of its application ecosystem – all of the positives that have made it so much fun and frustrating to both learn and use at the same time.
    Its too bad people viewed my earlier post as being arrogant and having an ‘attitude’ and being hypocritical. That wasn’t the intent of the post. Please re-read the last couple of paragraphs. Put your efforts into improving that ecosystem nstead of pining for the ‘good-old-days’ of 20 years ago. I’ve been doing this since 1968, and I really don’t want to go back 20 years to where no one even knew something like Linux even existed. We can do several things to deal with the Linux mess:
    – go to one of the ‘BSDs
    – go to ReactOS
    – go to Slackware or Gentoo
    – or –
    – get into helping projects where you want to see changes made
    – stop whining!
    Stop being slaves to the past – reminds me of Springsteen’s Glory Days.
    Be masters of your own environments. Don’t waste a lot of time with moaning and pity parties.


  15. Maybe it should be enough to tell you that Firefox on FreeBSD is still at version 27 (everywhere else it is at 35). Since Mozilla releases a version each 6 weeks, this means that FreeBSD is 48 weeks late – almost a year. And this happens with most packages. On the other BSD’s it’s even worse. The conclusion: everything 3rd party (i.e. not BSD proper) takes ages to be included in the official repository. Why? Because everybody’s on Linux distributions, now. Shame, but the BSD world didn’t know how to play its card in the early ’90, and that’s when it lost – 20 years ago. It’s just that they still haven’t realised it yet. They are like zombies: dead, but they still think they are alive. Sad, really.


    mns` Reply:

    A. you are obviously wrong. I don’t know if you did it on purpose, but the fact is, FreeBSD has a very updated userland tools:
    — FreeBSD 10.1
    % pkg search firefox


    A. Reply:

    Since this blog allows only a small level of imbricated reply levels, I shall answer my own comment: yes, I was mistaken, FreeBSD has a reasonably up-to-date 3rd party software base. I was mislead by the old versions found in http://ftp.freebsd.org/pub/FreeBSD/ports/packages/www/ . (And no, I didn’t “do it” on purpose…)


  16. Nope, you’re definitely in good company with these sentiments.

    I’ve run various incarnations of RH/Fedora on my laptops and workstations since ’04, with XFCE being my desktop of choice for the past 2-3.

    My experience is the same: annoying, niggling problems that used to be straightforward to isolate, diagnose and resolve by virtue of the platform’s simple, logical architecture have morphed into a morass that’s worse than the Windows Registry.


  17. I see this as being a couple of problems. I don’t have a solution for you, although I’d love to help. This is probably one of the following: the mess of libraries, programs, and config files most of which are completely incomprehensible necessary to get the linux desktop to the point where even an idiot can use it, and something is getting lost in translation. Or, it’s a bad implementation of systemd on debian’s part. To be honest, I run a mix of arch linux and ubuntu systems and I’ve never had anything like this. One the one occasion I did get a not authorized error, it was because I was trying to mount a drive with a user that couldn’t mount drives, because i’d set it up like that. This *is* a problem, but it seems to me that the linux community is more concerned that netflix, skype and steam work on their chosen platform than they are about making sure the documentation is well written, non geekish, and to be honest, in a format other than turse manpages. I share your pain, I just don’t have a solution


  18. FreeBSD has GNOME, KDE, xfce, and MATE too. Desktops there often lag a good bit behind Linux. GNOME 2 didn’t go away in favor of GNOME 3 fully until last month.

    The center of the “forced to” argument is that systemd was made into a package level requirement for too many things in many Linux distributions. You can’t really excise systemd for another init systems without taking out a lot of previously unrelated software. That’s more of a packaging dependency issue than anything else though. And it doesn’t exist in FreeBSD’s packaging yet.

    What will be interesting is if GNOME starts depending on systemd specific APIs too, so that the desktop environment itself won’t work without it. That would put a serious problem into FreeBSD continuing to use it.


  19. I’m sort of in the same boat in that I’ve felt control and understanding of my computer being taken away from me. At some point I got enough and simply threw out all the parts I didn’t know I need, in particular all this *kit junk. Fortunately for me, I’ve always felt more comfortable doing basic tasks on command line, so this didn’t require any major adjustment to my habits.

    For the mount problem my solution was to write my own minimalistic GUI for pmount. You can find it at http://git.tdb.fi/?p=pmount-gui.git;a=summary in case you’d like to give it a try.


  20. I’ll add another aspect to this, one that no one seems to have covered. That of accessibility. ALthough sometimes I find linux frustrating, usually only when the desktops try to be so brain dead simple to use they take out features I personally liked (cough, gnome 3) it fits me. I’ve tried various incarnations of windows and while they’re pretty accessible, there’s one aspect that linux has them beat, hands down. LInux has live disks with a screen reader available on the disks, allowing for independent installation, repair, etc. WIndows, even as of the latest 10 preview still does not have this. I suspect this is because MS is catering to the blind software companies who profit from this immensely. GOing As a blind person who depends on linux every day, that’s extremely important. back to linux being complex, yes, it is more complex. I think this is mostly do to the glue needed to make things like mounting of flash drives, dvd’s, etc work without root permission. Remember, none of this stuff is supported in x itself, so a lot of this has to be done outside of x and then somehow presented to the user. I really, really hope once wayland, mir, whichever takes over from x, that this stuff gets sorted out and documented. IN fact, I’ll go so far as to say i’d be willing to help with documentation, as that’s something I’m good at. i can’t code though. TO sum up, even though linux has it’s own issues, I’ll continue to use it. The blind community can be extremely frustrating sometimes, particularly if you’re trying to get some coders involved, but shrug. I think linux will eventually figure all this out.


  21. Have used debian for about 6 years and just wanted to voice my frustration at the current direction it’s heading!

    I think one of the reasons many of us gravitated toward linux in the first place, was the level of control that we could exercise over every part of the system.

    I didn’t have a clue how to configure things when i started, but because there was a consistency to how things were configured and the configuration was all handed in plain text i could fairly easily figure it out.

    The way it is now i feel like i have very little control over the underlying system. And making changes often becomes a ordeal that doesn’t always end with success.

    I don’t love or hate systemd, what i do have a problem with is the removal/obscuring of my ability to have fine grained control over my own system!

    I think systemd goals as i understand them were great, but it’s implementation is another matter.

    The continuing inclusion of necessary system function into systemd is wrong headed, it discourage development of alternatives to those individual components and it leaves the design decisions to a small group of developers. Sounds like “embrace, extend, and extinguish”.

    It will slow innovation and development of alternative designs of core system components and it removes many of the reasons that we chose linux in the first place.


  22. completely agree. I’m not frustrated with any particular part of linux, save it’s accessibility, which can seem like pulling teeth at times. I’m still positive though. If systemd, gnome, any part of linux really becomes a problem the community will do what it always does and develop an alternative. The big distros may be slow to adopt it, but if it gets enough adoption they’ll switch over. They will, that is, if they can stop playing steam and skyping long enough to realize there’s a need. I’m not trying to go off topic here, but I really think there’s a new crop of linux users who care less about the philosophy of open source and the infinite, or nearly so, customizeability of the system then pretty guis, microsoft office, steam, and all those various closed source packages that allow windows to dominate. This might be why the system is starting to veer away from it’s unix roots. No proof whatsoever, just what I think. I’m also not trying to start any flame wars. Can anyone give me a pointer on what might help this be less painful? Is more documentation needed? if so, where? What needs better documentation. I’m willing to subscribe to whatever mailing lists, go onto whatever irc channels, etc is necessary to start fixing this. I will not, however, start advocating for systemd’s removal, replacement, etc just yet. I happen to like systemd, but I will try my best to help if I can. Is it polkit? udev? udisks? If so, what needs documenting. Do people just need some kind of insite on how the system works? Or is there some kind of radical overhall of the entire system needed?


    Nate Bargmann Reply:

    I’m not even certain that this is borne of chasing Microsoft any more. Rather, it seems as though there is a certain percentage of the developers that are infatuated with all things Apple. Now, gathering inspiration is nothing new, it has gone on all the way through the history of software development. However, there is also the point of “eating our own dog food” whereby developers use the code they write. Is that the case any more?

    I ask as following blogs and Planet Debian for years I have read much commentary of the number of conference attendees using a Mac Book. I’ve also read of developers (even Debian developers) that proclaim their love of OS/X and are using it full time because of some shortcoming perceived or otherwise, of the free desktop. To this country boy that seems like a self fulfilling prophesy. In other words, when the developers can’t stand what they have written, how logical is it that the rest of us should “suck it up”?

    This is a problem and only seems to have manifested itself within the past five years or so, about the time frame that a lot of things started changing. Prior to that time the developers of desktop environments were proudly using their code at conferences and so on and now it seems as though they only show their code running inside an emulator on OS/X. As I have a difficult time believing that adherents to the Free Software philosophy would suddenly be smitten by OS/X and head en masse to a proprietary system, something else must explain what happened within the past five years.

    No, I am not being conspiratorial here. It seems to me that a lot of commercial development came to bear on the free desktop and I suspect a lot of developers were given tasks as part of their jobs. I wonder whether these developers really understood what the Free Software community valued as it seems that only within the past several years that I have read articles that the community is hard to work with or that the community is the problem. Could this explain why a lot of the infrastructure the community was comfortable with and supported has suddenly been declared obsolete and pushed aside?

    As John alludes in his followup article, pretty much everything that this new infrastructure is supposed to allow already worked in Wheezy. Wheezy was mostly built around the community supported infrastructure and when using Xfce everything worked quite well, at least for me. I’m not certain what, exactly, has been “fixed” between then and now, but I know that my Sid desktop systems seem considerably more complex today than a few years ago. If the developers were using the free desktop for everything they do on and off the clock, would it be this way today?


    Digi_owl Reply:

    I am tempted to call it the onset of the cloud monkeys.

    I see one symptom of it in how everything is to be solved with containers, including “dependency hell”…


  23. @kendell that’s not the point. the problem is that the average quality of open source in the Linux ecosystem is decreasing.
    I see it more as a cultural problem, as it is more socially acceptable to release crap.

    Regarding this specific XFCE mounting issue: somewhere, some day, a sloppy developer added the dependency to udisks and “forgot” to document it.

    How can this even pass any form of testing, I wonder? Simple: because no testing is being made whatsoever.


    sepp Reply:

    The problem with testing in this case is the “combinatorial explosion of the test matrix”. I.e. for every part of the system there are a number of alternatives (alternative implementations or different versions of the same implementation), and testing every possible combination is simply impossible.
    To solve this, one could either reduce the number of alternatives, which would violate the “Linux is about choice”-meme, or increase collaboration to get stable interfaces between the components.


  24. I don’t think I agree completely regarding the ‘quality’ of open source software. Some of it is great, some of it is crap. Same goes for the proprietary software. As documentation goes that was an oversite that should’ve been fixed. If a dependency on udisks was added, this should’ve been documented and added to … what is that file in the debian package … control? I’m sorry, I don’t know debian that well. I proudly use linux every day. I happen to like the gnome desktop, as well as mate. xfce, cinnamon, kde, are not usable by me because their accessibility is next to nill right now. Rather than excoriating the community for being lazy and not testing their software enough, I’m just glad I have access to linux at all. I’m confident that if the issue is bad enough it will get fixed. There are jerks in every community. Ther are lazy people. There are uncooperative developers. But just because people like that exist, that does *not* mean that linux as a hole is crap, or that the quality of linux is going down hill. SOme parts of it undoubtedly are, and those parts need a kick in the teeth. WHoever said that if you don’t use the code you write, and instead use a compeditors product or hardware, I couldn’t agree more. I used to see this all the time when I’d go into a cell phone store (this was when I used sprint). i’d constantly ask the sales people for help, usually something along the lines of does this phone talk? I’d get answers such as, “shrug, what do I know, I use at&t, verizen, etc.” My response was always, if you don’t even use the stuff you’re trying to sell to me, how am I supposed to feel confident in it’s quality? THis goes for linux as it does for anything else. I don’t think this necessarily means linux isn’t of high quality, or that it’s doomed. I get the feeling from some of these commenting here that some people think it is. Linux depends, and always has, on it’s community. If that community chooses instead to use bsd, apple, windows, anything other than linux, it will suffer. THings will fall through the cracks. This is not to say you cannot duel boot, but if the people who promote linux don’t use linux every day, instead using mac, windows, whatever because it ‘works better’ or does something else for them, that does not speak very well for their advocacy. I can’t trust someone who says one thing but does another. I’ve rambled enough, I’ll shut up now.


  25. john, since you’re the one who published this blog entry, what can I do to help? I don’t know if I can fix all the issues you’re having but I can do my best. Is it documentation that’s needed? Understanding on how the system works. Give me some pointers and I’ll keep you updated on my progress by email, if I can find your email address. Or comments here if you don’t feel comfortable giving me your email address.


    John Goerzen Reply:

    Hi Kendell,

    That would be *great*! I think our greatest needs are:

    1) Documentation for how all these pieces fit together. Tracing from, say, Digikam back through to /dev/sdc and where all the configuration bits come in to play would be immensely helpful. Ditto for network-manager, the keyring stuff, suspending, etc.

    2) Practical examples of how to change things or override defaults without doing so in /usr where changes are prone to be overwritten by distro upgrades.

    3) A sort of “reverse analysis” – figuring out who/what has access to what on a system for the purposes of securing it.

    My email address is available at http://www.complete.org/JohnGoerzen and you are welcome to use it.


  26. got it. I’ll start pinging the irc channels now. I have some vague ideas where to start. I think most of the automounting stuff gets it’s authentication from either pam or policykit. Both use rule files that can be changed or added to. Digicam probably polls udev to make sure what you’ve inserted is a camera, then mounts it via fuse. I’m still not sure where udisks comes in. I’ll look for documentation on google, and if I don’t find any there I’ll start looking for irc channels and mailing lists


  27. The problem with Linux and almost all opensource Linux based software is the devs, only design things for themselves. The creative process involves a major amount of epeen. “what your not as smart as me, and can’t see what is plain as day to me what this program does and how it works?”

    You see in in almost all the Linux releases and large open source linux based applications like Drupal.


  28. AFAIK, Debian Jessie still uses an old version of polkit (<0.106) that doesn't store its configuration in Javascript. So this exact solution wouldn't work there (though it wouldn't be hard to convert it to the old pkla-format).


Leave a Reply

Your email address will not be published. Required fields are marked *