You missed the point completely. The point is that people complaining that Debian changed the default init system and that there is currently a GR to make it mandatory for upstream developers to make sure their packages run at least systemd plus something else. Which means that the upstream devs have to work to support something they don't have any interest in doing so. At that only because some complain that they should be able to replace systemd for whatever reason. That is akin to demanding that upstream devs must make sure that their packages compile with gcc + something else, or that their software must run with Linux+ something else, because some want to replace the Linux kernel.
I did not wrote that some people replaced sysvinit, or patched it or whatever. The current issue is the GR and the decision of the TC.
Yet nobody has proposed a GR forcing support for kFreeBSD or the Hurd; the people working on them have simply *done the work*, and in some cases successfully convinced others to do the same.
(And analogously, non-Linux kernels such as FreeBSD often have substantial shim layers providing Linux APIs for the purposes of porting software.)
I'm interested in Linux, and the other topics don't interest me so much. And I was reading the emails on debian-voting about the TC, and the GR proposed by Ian.
Whatever. I use what works, is simple and well documented. That is my only idée fixe, if you want to call it that. This "not UNIX", "written my Lennart", "is pushed on users" is not coming from me.
So, your issue is that systemd does not support a feature you relied on. That is a valid complain, but hardly limited to systemd. Even the Linux kernel throws out old drivers and changes the kernel APIs from time to time, and userland applications change their features much more frequently.
The original topic was "updated to jessie and installed systemd by default, had to roll back VM and pin systemd. Fuck Debian." If ruir had a substantial argument like you have, I wouldn't have replied at all, but modded +1 informative.
I think this whole thing would be a non-issue if you could swap out systemd with another system and still have everything function easily.
There was no complains with sysvinit in the same regard. People just used sysvinit because it was the default one and nobody complained "but I want to replace sysvinit with xxxx and still have everything function easily".
Now the default one is systemd, who cares. And if developers of upstream projects like Gnome find systemd so helpful that they use systemd's features and thus Gnome can't be used anymore with some inferior system, well, that is called progress. The replacement systems must then keep up and offer similar features.
Linux should go always with the technically superior, and always went with that in the past. That is the main reasons I like open source and Linux, there is no politics but only the technically superior wins. Linus build a monolithic kernel because it is technically superior, Xfree was replaced with Xorg, and will be replaced with Wayland. The sound stack was replaced with PulseAudio. And so on.
Since only 2011 Debian supports a different kernel then Linux (kFreeBSD) (Debian was founded in 1993), so it does not make any sense to argue that Debian should support other kernels for a "diversity of systems". I don't think there is any commitment of a diversity of init systems. By your "reasonable definition" nothing on my Linux desktop is modular. IPC messages are the highest possible abstraction, then comes dynamic libraries, then static libraries.
As far as I understand uselessd is not using Kdbus, which is the future default IPC protocol of the Linux kernel. If you need the features of uselessd, go for it. But it have nothing to do with your argument.
One side wages an "ideological and cultural war", the other made their decision on technical terms. Only one side pushes ideological arguments like "systemd is not the UNIX way" or "systemd is pushed by Pottering on users" or idiotic arguments like "PID 1 should not be complicated like systemd"
Poettering has been increasingly vocal that he wants systemd to be more than an init system.
Who fucking cares? systemd is modular, Debian can just pull in the init stuff of systemd. How is that relevant that the guy who wrote a software peace that Debian wants to use want to add new features?
Installing GIMP, for example, will pull in systemd libs.
Some optional dependency of GIMP pulls systemd-libs. And who cares anyway, if it's just a library? You know that GIMP depends on a bunch of libs one more or one less who cares. A systemd-lib is not systemd.
Your premises are not sound. - "objectionable approach" is just irrelevant, because it's subjective. Many comments here are just personal attacks on LP, so, no matter which approach systemd would have chose, it would always be "objectionable approach". - greater flexibility and hackability that's just a hypothetical. systemd already have great flexibility and is open source for hacks
In any case, this discussion is irrelevant. The systemd devs chose their approach, and you have nothing, just meaningless critiques. How about you start to develop your solution, and then ask the Debian TC to change the init system to use your solution?
That question is easy. Because the TC of Debian voted and decided to change the default init system to systemd. If you start your own distribution and set up your own organization, then you can decide which default init system you are going to use.
*breaking news* everything on a computer is binary. As if there is any difference to use journald or grep or emacs to read the log. journald can even be statically linked without systemd, so that it can be put in BusyBox. In addition, journald can put advanced features in the log, like signing, crypt, etc, and it have nice user interface, like journalctl --since "20 min ago" And finally, you still can use syslog or rsyslog with journald.
Compatibility with a classic syslog implementation can be provided by letting systemd forward all messages via the socket/run/systemd/journal/syslog. To make the syslog daemon work with the journal, it has to bind to this socket instead of/dev/log
Fine, I though we are talking of a commercial product (" 3. material created or produced and viewed in terms of potential sales"). So, how exactly is Red Hat benefiting by giving away a free init system that their direct competition can use? As for example SUSE Linux GmbH, Mandriva or Debian.
In this threat you did not point to anything that is "anti-UNIX". You just complained why systemd must be PID 1 and thus forces people to learn new stuff. Or is the "UNIX way" that we must do everything with shell scripts?
How is systemd a product? How is Red Hat going to make money with it? Red Hat is a company, they don't write software just for the fun of it. Your bias against Red Hat developers can't be more obvious. What do I care about some whining Slashdot comments? I have yet to read any technical argument against systemd, usually it's boils down to either of the following - not "Unix" - it's new - binary logs - it's from Lennart Poettering And those are just idiotic arguments.
Systemd is not a product of RedHat employees, Red Hat Linux is. And RedHat employees feel the need to develop systemd for the success of Red Hat Linux. And Debian feels the need to replace sysvinit with a modern init system. It was either systemd or upstart, and upstart was not ready.
Slashdot Mirror
You missed the point completely. The point is that people complaining that Debian changed the default init system and that there is currently a GR to make it mandatory for upstream developers to make sure their packages run at least systemd plus something else. Which means that the upstream devs have to work to support something they don't have any interest in doing so. At that only because some complain that they should be able to replace systemd for whatever reason. That is akin to demanding that upstream devs must make sure that their packages compile with gcc + something else, or that their software must run with Linux+ something else, because some want to replace the Linux kernel.
I did not wrote that some people replaced sysvinit, or patched it or whatever. The current issue is the GR and the decision of the TC.
See for example
https://lists.debian.org/debia...
Yet nobody has proposed a
GR forcing support for kFreeBSD or the Hurd; the people working on them
have simply *done the work*, and in some cases successfully convinced
others to do the same.
(And analogously, non-Linux kernels such as FreeBSD often have
substantial shim layers providing Linux APIs for the purposes of porting
software.)
I'm interested in Linux, and the other topics don't interest me so much.
And I was reading the emails on debian-voting about the TC, and the GR proposed by Ian.
Whatever. I use what works, is simple and well documented. That is my only idée fixe, if you want to call it that.
This "not UNIX", "written my Lennart", "is pushed on users" is not coming from me.
Well, many people disagree with you. That's life.
systemd is open source and free. If you are so concerned then you can just convert all your logs to text.
So, your issue is that systemd does not support a feature you relied on. That is a valid complain, but hardly limited to systemd. Even the Linux kernel throws out old drivers and changes the kernel APIs from time to time, and userland applications change their features much more frequently.
The original topic was "updated to jessie and installed systemd by default, had to roll back VM and pin systemd. Fuck Debian."
If ruir had a substantial argument like you have, I wouldn't have replied at all, but modded +1 informative.
I think this whole thing would be a non-issue if you could swap out systemd with another system and still have everything function easily.
There was no complains with sysvinit in the same regard. People just used sysvinit because it was the default one and nobody complained "but I want to replace sysvinit with xxxx and still have everything function easily".
Now the default one is systemd, who cares. And if developers of upstream projects like Gnome find systemd so helpful that they use systemd's features and thus Gnome can't be used anymore with some inferior system, well, that is called progress. The replacement systems must then keep up and offer similar features.
Linux should go always with the technically superior, and always went with that in the past. That is the main reasons I like open source and Linux, there is no politics but only the technically superior wins. Linus build a monolithic kernel because it is technically superior, Xfree was replaced with Xorg, and will be replaced with Wayland. The sound stack was replaced with PulseAudio. And so on.
What do you think less or more are? special tools to read files.
Also journalctl -b | less works just fine.
That's just idiotic.
Since only 2011 Debian supports a different kernel then Linux (kFreeBSD) (Debian was founded in 1993), so it does not make any sense to argue that Debian should support other kernels for a "diversity of systems". I don't think there is any commitment of a diversity of init systems.
By your "reasonable definition" nothing on my Linux desktop is modular. IPC messages are the highest possible abstraction, then comes dynamic libraries, then static libraries.
What rest? Systemd core is just systemd, journald, networkd and logind.
http://en.wikipedia.org/wiki/F...
As far as I understand uselessd is not using Kdbus, which is the future default IPC protocol of the Linux kernel.
If you need the features of uselessd, go for it. But it have nothing to do with your argument.
I have no relationship. Just using happy Fedora 20 with systemd. I just want to hear a genuine error of systemd.
And what problems did you had?
Sysvinit scripts run just fine under systemd, syslog runs fine, too.
One side wages an "ideological and cultural war", the other made their decision on technical terms.
Only one side pushes ideological arguments like "systemd is not the UNIX way" or "systemd is pushed by Pottering on users" or idiotic arguments like "PID 1 should not be complicated like systemd"
Poettering has been increasingly vocal that he wants systemd to be more than an init system.
Who fucking cares? systemd is modular, Debian can just pull in the init stuff of systemd. How is that relevant that the guy who wrote a software peace that Debian wants to use want to add new features?
Installing GIMP, for example, will pull in systemd libs.
Some optional dependency of GIMP pulls systemd-libs. And who cares anyway, if it's just a library? You know that GIMP depends on a bunch of libs one more or one less who cares. A systemd-lib is not systemd.
What commercials? oO
LOL good one.
Your premises are not sound.
- "objectionable approach" is just irrelevant, because it's subjective. Many comments here are just personal attacks on LP, so, no matter which approach systemd would have chose, it would always be "objectionable approach".
- greater flexibility and hackability
that's just a hypothetical. systemd already have great flexibility and is open source for hacks
In any case, this discussion is irrelevant. The systemd devs chose their approach, and you have nothing, just meaningless critiques.
How about you start to develop your solution, and then ask the Debian TC to change the init system to use your solution?
That question is easy. Because the TC of Debian voted and decided to change the default init system to systemd.
If you start your own distribution and set up your own organization, then you can decide which default init system you are going to use.
*breaking news* everything on a computer is binary.
As if there is any difference to use journald or grep or emacs to read the log. journald can even be statically linked without systemd, so that it can be put in BusyBox.
In addition, journald can put advanced features in the log, like signing, crypt, etc, and it have nice user interface, like journalctl --since "20 min ago"
And finally, you still can use syslog or rsyslog with journald.
Compatibility with a classic syslog implementation can be provided by letting systemd forward all messages via the socket /run/systemd/journal/syslog. To make the syslog daemon work with the journal, it has to bind to this socket instead of /dev/log
https://wiki.archlinux.org/ind...
Fine, I though we are talking of a commercial product (" 3. material created or produced and viewed in terms of potential sales"). So, how exactly is Red Hat benefiting by giving away a free init system that their direct competition can use? As for example SUSE Linux GmbH, Mandriva or Debian.
In this threat you did not point to anything that is "anti-UNIX".
You just complained why systemd must be PID 1 and thus forces people to learn new stuff.
Or is the "UNIX way" that we must do everything with shell scripts?
Systemd is modular and open source. I don't know what issues you have with it.
How is systemd a product? How is Red Hat going to make money with it?
Red Hat is a company, they don't write software just for the fun of it.
Your bias against Red Hat developers can't be more obvious.
What do I care about some whining Slashdot comments? I have yet to read any technical argument against systemd, usually it's boils down to either of the following
- not "Unix"
- it's new
- binary logs
- it's from Lennart Poettering
And those are just idiotic arguments.
Systemd is not a product of RedHat employees, Red Hat Linux is. And RedHat employees feel the need to develop systemd for the success of Red Hat Linux.
And Debian feels the need to replace sysvinit with a modern init system. It was either systemd or upstart, and upstart was not ready.
Because systemd "respects the nature of Unix".