I Switched (Back) Over To Fedora As My Main OS & It's Going Great!

[cocklover]

that is your hope and hope cannot be opinion.... EVER

arch started using it and lost almost 0 popularity. opensuse started using it and look... same result. fedora uses it from beginning... holly shit, same result. ubuntu/debian plans on using it and you don't see people flocking away en mass on that reason. whole thing is reduced to few vocal people that make 0 difference

and there is one important thing at hating some tech. understanding why, which you don't

NO. I was using Arch linux when they make the switch. So many post of broken systems, merging on one big thread and common. People ranting on mailist, trolling, etc. It was a hard time. People banned for asking for alternatives. Arch lost his rc central file and moved to 6 or more files to make config, they lost the kiss in favor of the distro maintainers and not users... wich has to deal with systemd anyways

Again for the majority (that doesn't have to include you) its just an implementation detail. Those user only notice that there system boots up a bit faster and that's basically it.

NO. You have to deal with systemd always is NOT only a init system, example if you want to power off/reboot the system from the terminal(We can't even do that on upstart cause of systemd monopoly breaking things on alternatives), if you want to change the clock from utc to localtime, if you want to start/enable/disable some service example apache, mysql, or even some upnp servers, if you have some bug on some program example postgresql you will be adressed to read systemd binary logs... you have to learn about journalctl and forget about /var/logs, if you want to load some kernel modules example virtual box module you have to deal with systemd, SO NO BOY, systemd IS NOT INVISIBLE FOR THE USER systemd.

[hooluupog]

Well, that's good. Personally I think Canonical should be totally boycotted. Anybody not blinded can see them for what they are. It's good to see you using a distribution where they actually contribute back to upstream projects. (upstream projects that they didn't create) (Ubuntu only reliably contributes to projects they maintain, and that's bad for everyone, including themselves.)

As a regular user who just need an out of the box linux desktop system to replace osx and windows, I don't care about this. Linux desktop sucks so many years and ubuntu is the least bad choice from a novice perspective.

[asdfblah]

I don't get why people uses this or that distro for the DE. Are you people so shortsighted that you don't think you can install this or that DE in every single distro (through the main repos and/or external repos)? Distro-hopping just doesn't make sense to me.
I use the distro and the DE I use (Kubuntu) because of the package management, the packages themselves and the overall configurability. I've tried many in the past, but so far, I prefer KDE. I gotta admit though, I don't care much about looks, I care about functionality and ease of use.

[Michael_S]

that is your hope and hope cannot be opinion.... EVER

arch started using it and lost almost 0 popularity. opensuse started using it and look... same result. fedora uses it from beginning... holly shit, same result. ubuntu/debian plans on using it and you don't see people flocking away en mass on that reason. whole thing is reduced to few vocal people that make 0 difference

and there is one important thing at hating some tech. understanding why, which you don't

Common systemd criticisms and my responses:

1. "I'm already comfortable with init or upstart and I don't want to learn anything different no matter what it is, who wrote it, what benefits it has, or how it works."
-- I consider this the weakest argument, it's the same argument made by people that will never give up Windows XP, or OS X, or a typewriter because they hate learning new things. But I'm not trying to group everyone who dislikes systemd in this category.

2. "A core piece of Unix philosophy is building small components that do one thing well. Systemd has too many pieces and too many jobs locked into one piece of software."
-- While this principle is useful in general, a few targeted exceptions have been successful. The Linux kernel is monolithic while the older HURD project of smaller pieces, which is more true to the Unix philosophy, has not reached production-ready yet. ZFS takes features normally used outside of a filesystem (snapshots, embedded error-checking, volumes) into it, and it's wildly popular - and the design of btrfs attempts to copy most of the good features in ZFS. Perl fills many roles, and one of those is as a replacement of sed and awk - bigger, more complicated, more features, but ultimately worth the complexity and the learning curve because it can do more (though to be nitpicky this comparison isn't totally fair, since you can still have sed, awk, and Perl alongside each other).

3. "Adoption of systemd is becoming invasive, in the sense that it now includes dbus and has become a dependency for newer versions of GNOME among other projects."
-- I think this is a fair point. Poettering makes the point on the mailing lists that most of the things that depend upon systemd really depend upon an interface, and anyone can implement that interface without using systemd. That's true, but it does make it more work for the maintainers. Now if you have init, or Upstart, or runit, or an init system you wrote yourself, you have to do extra work to make GNOME work. However, I don't think this is enough to justify rejecting systemd.

4. "Systemd is monolithic. You have to include all of it."
-- Not true. You can compile different components as part of it and leave others out.

5. "Systemd requires all of your init code to be written in C."
-- Not true. You can launch shell scripts with systemd. You can disable systemd modules written in C and replace them with shell scripts.

6. "Systemd is large. With an init system with 550,000+ lines of code, there is more room for errors and security code than in other init systems."
-- It's a false comparison. SysV init has orders of magnitude fewer lines of code in core init, but it includes many hundreds of thousands of lines of shell scripts plus the 100,000+ lines of code in bash itself and of course systemd includes features that Sysv init doesn't provide.

7. "There are ties between the systemd version and the kernel version, and you can't upgrade them independently of each other. Likewise, systemd is tied to the Linux kernel and cannot be used in the open source *BSD Unixes or OpenSolaris or any other operating system."
-- I think that's the most solid criticism. I can understand allowing some components to be tied into init that traditionally were separate. But I think life would be easier for kernel developers, systemd developers, distribution maintainers, and users if someone could slap systemd 218 on a Linux 3.10 kernel or systemd 204 on a 3.18 kernel without thinking about it. However, one of the primary goals of systemd is to take advantage of Linux kernel APIs that are specific to Linux and not part of POSIX. For example, cgroups are used as part of controlling and monitoring services. That gives systemd a lot of flexibility and control, but cgroups are unique to the Linux kernel.

8. "There are showstopper systemd bugs reported."
-- Yes. To my knowledge, they are exceedingly rare. I don't mean to downplay their critical nature when they occur. But to use these bugs as an excuse to ban systemd would be as silly as using the Shellshock bug as a reason to ban the bash shell or Heartbleed as a reason to ban OpenSSL. A high frequency of defects is a reason to avoid using something. To my knowledge, the systemd defect rate is quite low. I've had startup problems with SysV init, nobody suggested we ban that. And yes, I realize Poettering made a lot of enemies with Pulseaudio years before. But there were six years between when pulseaudio started and when systemd started, and it's been five years since systemd started. A developer can learn a lot in that time. I'm somewhere in the neighborhood of competent as a software developer now, I was terrible in 2004. I also haven't had a problem with pulseaudio since... 2007? 2006?

[halo9en]

whee, another systemd thread

[Luke_Wolf]

7. "There are ties between the systemd version and the kernel version, and you can't upgrade them independently of each other. Likewise, systemd is tied to the Linux kernel and cannot be used in the open source *BSD Unixes or OpenSolaris or any other operating system."
-- I think that's the most solid criticism. I can understand allowing some components to be tied into init that traditionally were separate. But I think life would be easier for kernel developers, systemd developers, distribution maintainers, and users if someone could slap systemd 218 on a Linux 3.10 kernel or systemd 204 on a 3.18 kernel without thinking about it. However, one of the primary goals of systemd is to take advantage of Linux kernel APIs that are specific to Linux and not part of POSIX. For example, cgroups are used as part of controlling and monitoring services. That gives systemd a lot of flexibility and control, but cgroups are unique to the Linux kernel.

There's actually an additional problem with this argument. No one else important other than linux was actually using SysV, the BSDs used their own BSDInit from the beginning, Solaris uses SMF, and OS X uses launchd.

[justmy2cents]

NO. I was using Arch linux when they make the switch. So many post of broken systems, merging on one big thread and common. People ranting on mailist, trolling, etc. It was a hard time. People banned for asking for alternatives. Arch lost his rc central file and moved to 6 or more files to make config, they lost the kiss in favor of the distro maintainers and not users... wich has to deal with systemd anyways

yet, everyone i knew who used arch still uses it, except 2 or 3 distro hoppers who never stick with anything anyway.
and let me clue you on 2 things
- when person is unhappy there is 1/50 chance he will whine and scream, when person is happy there is 1/5000 he will stand up and praise. that's just human factor. or do you see arch dying? i don't, if anything damn thing is getting more and more popular
- each new tech has problems when it is in start phase of enabling. there are simply too many factors that one can forget to check. if you think otherwise, try some big scale change for your self

NO. You have to deal with systemd always is NOT only a init system, example if you want to power off/reboot the system from the terminal(We can't even do that on upstart cause of systemd monopoly breaking things on alternatives), if you want to change the clock from utc to localtime, if you want to start/enable/disable some service example apache, mysql, or even some upnp servers, if you have some bug on some program example postgresql you will be adressed to read systemd binary logs... you have to learn about journalctl and forget about /var/logs, if you want to load some kernel modules example virtual box module you have to deal with systemd, SO NO BOY, systemd IS NOT INVISIBLE FOR THE USER systemd.

what do you mean with power off/reboot from terminal? i see absolutely no change here. same commands still work

switching time display was always clusterfuck on linux. each distro had different approach. but, i cant remember one time i would need it

enabling and disabling services? i don't know, but on fedora they did awesome job at it. all old commands still work and when you execute for example "chkconfig httpd on", not only it does what it is supposed to, it also writes out how same thing is done on systemd natively

i have to deal with both kind of logs. old and new on my rhel installs. and there is no difference for me. and if it wouldn't be i could install syslogd whenever i wanted. it's not like it takes over kernel facilities. but, i really got used to journal and i like it. but, in most cases i can live with "systemctl status servicename" since it provides last lines of log which usually contain error

i think your main problem is "you have to learn"

[justmy2cents]

Common systemd criticisms and my responses:

1. "I'm already comfortable with init or upstart and I don't want to learn anything different no matter what it is, who wrote it, what benefits it has, or how it works."
-- I consider this the weakest argument, it's the same argument made by people that will never give up Windows XP, or OS X, or a typewriter because they hate learning new things. But I'm not trying to group everyone who dislikes systemd in this category.

2. "A core piece of Unix philosophy is building small components that do one thing well. Systemd has too many pieces and too many jobs locked into one piece of software."
-- While this principle is useful in general, a few targeted exceptions have been successful. The Linux kernel is monolithic while the older HURD project of smaller pieces, which is more true to the Unix philosophy, has not reached production-ready yet. ZFS takes features normally used outside of a filesystem (snapshots, embedded error-checking, volumes) into it, and it's wildly popular - and the design of btrfs attempts to copy most of the good features in ZFS. Perl fills many roles, and one of those is as a replacement of sed and awk - bigger, more complicated, more features, but ultimately worth the complexity and the learning curve because it can do more (though to be nitpicky this comparison isn't totally fair, since you can still have sed, awk, and Perl alongside each other).

3. "Adoption of systemd is becoming invasive, in the sense that it now includes dbus and has become a dependency for newer versions of GNOME among other projects."
-- I think this is a fair point. Poettering makes the point on the mailing lists that most of the things that depend upon systemd really depend upon an interface, and anyone can implement that interface without using systemd. That's true, but it does make it more work for the maintainers. Now if you have init, or Upstart, or runit, or an init system you wrote yourself, you have to do extra work to make GNOME work. However, I don't think this is enough to justify rejecting systemd.

4. "Systemd is monolithic. You have to include all of it."
-- Not true. You can compile different components as part of it and leave others out.

5. "Systemd requires all of your init code to be written in C."
-- Not true. You can launch shell scripts with systemd. You can disable systemd modules written in C and replace them with shell scripts.

6. "Systemd is large. With an init system with 550,000+ lines of code, there is more room for errors and security code than in other init systems."
-- It's a false comparison. SysV init has orders of magnitude fewer lines of code in core init, but it includes many hundreds of thousands of lines of shell scripts plus the 100,000+ lines of code in bash itself and of course systemd includes features that Sysv init doesn't provide.

7. "There are ties between the systemd version and the kernel version, and you can't upgrade them independently of each other. Likewise, systemd is tied to the Linux kernel and cannot be used in the open source *BSD Unixes or OpenSolaris or any other operating system."
-- I think that's the most solid criticism. I can understand allowing some components to be tied into init that traditionally were separate. But I think life would be easier for kernel developers, systemd developers, distribution maintainers, and users if someone could slap systemd 218 on a Linux 3.10 kernel or systemd 204 on a 3.18 kernel without thinking about it. However, one of the primary goals of systemd is to take advantage of Linux kernel APIs that are specific to Linux and not part of POSIX. For example, cgroups are used as part of controlling and monitoring services. That gives systemd a lot of flexibility and control, but cgroups are unique to the Linux kernel.

8. "There are showstopper systemd bugs reported."
-- Yes. To my knowledge, they are exceedingly rare. I don't mean to downplay their critical nature when they occur. But to use these bugs as an excuse to ban systemd would be as silly as using the Shellshock bug as a reason to ban the bash shell or Heartbleed as a reason to ban OpenSSL. A high frequency of defects is a reason to avoid using something. To my knowledge, the systemd defect rate is quite low. I've had startup problems with SysV init, nobody suggested we ban that. And yes, I realize Poettering made a lot of enemies with Pulseaudio years before. But there were six years between when pulseaudio started and when systemd started, and it's been five years since systemd started. A developer can learn a lot in that time. I'm somewhere in the neighborhood of competent as a software developer now, I was terrible in 2004. I also haven't had a problem with pulseaudio since... 2007? 2006?

lol, hope this was meant as joke or at least continuation of my argument since it was answer to me. i like systemd and use it ever since fedora implemented it as default. i wouldn't go back to old... EVER!

[jimbohale]

Fedora has systemd and gnome shell. 90 % of linux user hate both. And you don't have decent repos/video drivers/video codecs/decent package maneger like synaptic... and is a beta for red hat. So...

87% of all statistics are made up

[Alejandro Nova]

Until systemd breaks something -again-... Even though they may have no idea what caused it, they care. All they may know is that something is broken -again-. They definitely care.

I don't think so, they say "Ubuntu/whatever is broken".

AFAIK systemd packaging for Debian is severely lacking, and Ubuntu's implementation of systemd has suffered/is suffering because of packaging issues. On the other hand, Fedora's systemd implementation is overseen by Lennart Poettering himself: if something fails, Lennart and the upstream systemd dev team themselves will write the patch, upload it to Koji, help out with testing, and you'll get it on fedora-updates-testing in a matter of minutes. That's why the majority of systemd haters come from Debian/Ubuntu, while the majority of systemd lovers tend to come from Fedora, OpenSuSE, and other distros with systemd well implemented.