The abysmal state of GNU/Linux and a case against shared object libraries

Personal critique

This is in part difficult to write since I have been daily driving GNU/Linux for 20 years now, but I think it is necessary to be honest about this. How come GNU/Linux is worse than it was 10 years ago? This may very well be a subjective opinion, GNU/Linux being better in the past, or maybe I am looking at the situation with rose-tinted glasses.

A full disclaimer, this weekend my system that is pretty vanilla just decided to die after an update. And this was not a full version upgrade. Just a normal update that I do many times per week.

Sure, we now have PipeWire and Wayland. We enjoy many modern advances and yet, the practical use for me is worse than it was 10 years ago. Now all of a sudden, I can't rely on the system to be stable like it was. I don't remember the system bricking after an update, or the system becoming laggy after 10 days of uptime. This may be the issue with Fedora, though.

Over the years, I have daily driven many distributions. From Gentoo, Arch, Fedora to Ubuntu. My best memories were always with Debian. Just pure Debian always proved to be the most stable system. I never had issue or system breaking after an update. I can't say the same for Fedora.

From the get-go, I had issues. I have an Nvidia card and even booting presented issues on occasion. This never happened on other distributions, though they had their problems. Updating the system was basically an exercise in gambling. How come an operating system that boasts with the stability is so unstable? And this was not isolated to my main machine. This also happened on my X220 ThinkPad with Fedora on.

Shared dependencies were a mistake! There, I said it! I understand that disk space was limited back then and this was a legitimate constraint. But this has given me more grief than any other thing. I am all in for AppImages or something like that. I don't care if these images are 10x bigger. Disk space now is plenty, and they solve the issue with "libFlac.8.so is missing" and I have version 12 installed. Which comes with unnecessary symlinking, downloading of older versions and hoping that this will resolve the issue.

Now, the biggest apologist of GNU/Linux will never admit this and even saying something is wrong with this is considered a mortal sin. I, however, am not concerned with cultist behaviors. This is bullshit! Things should be better than 10 years ago, not worse. And I don't care how much lipstick you put on this pig. After more than 20 years of using Linux as my main system, I think I have earned a badge that gives me the right to say the truth.

Regardless of all this, I am still a massive fan. I still think GNU/Linux is probably the most unobtrusive operating system, bar none. But the complexity has gotten the best of it. It's bloated and too complicated at this point. Understandably, you can't have a modern operating system that competes with alternatives without sacrificing simplicity. But I still think that there is another way.

One of the best aspects of GNU/Linux must be outstanding package manager support. Nevertheless, they are essentially solving a problem that should have been solved and done with years ago. The number of gymnastics that happen in the background for you to install a software is just mind-boggling. The dependency graphs are insane. And Snaps and Flatpaks tried to solve some of these things, but until a distribution comes out that is completely devoid of shared dependencies, we will still live in this purgatory.

Let's compare these two distributions when it comes to packages sizes and shared object libraries and see how they fair.

Debian

root@debian: cat /etc/os-release

PRETTY_NAME="Debian GNU/Linux 12 (bookworm)"
NAME="Debian GNU/Linux"
VERSION_ID="12"
VERSION="12 (bookworm)"
VERSION_CODENAME=bookworm
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/"

Top 20 packages by size

dpkg-query -W --showformat='${Installed-Size}\t${Package}\n' | sort -nr | head -n 20 | awk '{size=$1;unit="KB"; if (size >= 1024) {size=size/1024; unit="MB"}; if (size >= 1024) {size=size/1024; unit="GB"}; printf "%.2f %s\t%s\n", size, unit, $2}'

This is more or less a fresh system that is being used daily for work.

Number of packages installed

dpkg -l | grep '^ii' | wc -l

2217

Number of shared object libraries on system (*.so)

Note: Some of them could be symlinks to each other so take this with a grain of salt.

find /lib /lib64 /usr/lib /usr/lib64 -follow -type f -name "*.so.*" | wc -l

5156

Shared objects sorted by size

find /lib /lib64 /usr/lib /usr/lib64 -type f -name "*.so.*" -exec du -h {} + | sort -rh | head -n 20

Fedora

root@fedora: cat /etc/os-release

NAME="Fedora Linux"
VERSION="36 (Workstation Edition)"
ID=fedora
VERSION_ID=36
VERSION_CODENAME=""
PLATFORM_ID="platform:f36"
PRETTY_NAME="Fedora Linux 36 (Workstation Edition)"
ANSI_COLOR="0;38;2;60;110;180"
LOGO=fedora-logo-icon
CPE_NAME="cpe:/o:fedoraproject:fedora:36"
HOME_URL="https://fedoraproject.org/"
DOCUMENTATION_URL="https://docs.fedoraproject.org/en-US/fedora/f36/system-administrators-guide/"
SUPPORT_URL="https://ask.fedoraproject.org/"
BUG_REPORT_URL="https://bugzilla.redhat.com/"
REDHAT_BUGZILLA_PRODUCT="Fedora"
REDHAT_BUGZILLA_PRODUCT_VERSION=36
REDHAT_SUPPORT_PRODUCT="Fedora"
REDHAT_SUPPORT_PRODUCT_VERSION=36
PRIVACY_POLICY_URL="https://fedoraproject.org/wiki/Legal:PrivacyPolicy"
SUPPORT_END=2023-05-16
VARIANT="Workstation Edition"
VARIANT_ID=workstation

Top 20 packages by size

rpm -qa --queryformat '%{SIZE}\t%{NAME}\n' | sort -rn | head -n 20 | awk '{size=$1;unit="B"; if (size >= 1024) {size=size/1024; unit="KB"}; if (size >= 1024) {size=size/1024; unit="MB"}; if (size >= 1024) {size=size/1024; unit="GB"}; printf "%.2f %s\t%s\n", size, unit, $2}'

What is interesting that the most of these packages are from a a system that are daily in use.

Number of packages installed

dnf list installed | tail -n +3 | wc -l

3484

Number of shared object libraries on system (*.so)

Note: Some of them could be symlinks to each other so take this with a grain of salt.

find /lib /lib64 /usr/lib /usr/lib64 -follow -type f -name "*.so.*" | wc -l

8894

Shared objects sorted by size

find /lib /lib64 /usr/lib /usr/lib64 -type f -name "*.so.*" -exec du -h {} + | sort -rh | head -n 20

Quick observation

/usr/lib/x86_64-linux-gnu/libjavascriptcoregtk-6.0.so.1.1.14
/usr/lib/x86_64-linux-gnu/libjavascriptcoregtk-4.1.so.0.4.14
/usr/lib/x86_64-linux-gnu/libjavascriptcoregtk-4.0.so.18.23.14

These three packages are probably used by three different applications, and nothing else needs them. I could be mistaken, but in any case they should just be packaged alongside the application that requires them and be done with it.

So, now what?

The main point is that, yes these packages can be big, but if we are honest, what would a couple of additional megabytes that would include shared object libraries actually do? Probably nothing! And it would make maintenance an entirely different game.

There is also this big elephant in the room, the users. They aren't concerned about package dependencies. They don't care if an application is 20 megabytes bigger. Nobody cares! But they certainly do care about borked systems and non-working dependencies and hunting for solutions for why libFlac was not found even though they have it installed (allthough slightly different version).

Operating systems should abstract these complexities away from the user. And I am not saying that the Linux kernel is at fault. I mean the operating system as a whole.

This is why I am a massive proponent of AppImages. Flatpaks and Snaps are OK, but they still have their own package management and the whole dependency problem because they try to optimize for space, and they sacrifice the simplicity.

It would be an interesting exercise to make a prototype distribution that does not rely on shared objects, but has everything packed in AppImages. Probably a foolish endeavor, but maybe worth looking into. I sense this kind of distribution would be highly unusable. Interesting how far we have gotten that this kind of distribution is almost not possible anymore.

The year of the GNU/Linux desktop? I have strong doubts. We are in a worse state than we were. Not only that, but the sheer amount of choice is paralyzing at this point. This is very relatable to The Paradox of Choice. The more options we have, the worse it gets. Wayland competing with X. So many window managers, you just get lost. So many choices. I have no idea if this is even salvageable, or something new must be made.

A quick shoutout to AppImageHub. If possible, I continually try to find applications there and take care of updating them myself. I don't need hand holding or a constant up-to date system. I just want my system to be stable and when some application has gotten some significant new features I can download that myself. It's about the choice and not being forced into this churn that requires constant updating and keeping up with things. At this point, using GNU/Linux is more like a second job, and I was so stoked when this was not a case anymore in the past. This is why I feel like the last 10 years were a regression disguised as progress.

Some interesting talks and videos

• Jonathan Blow on how an operating system should work
• The Thirty Million Line Problem by Casey Muratori
• GNU is Bloated! by Luke Smith