| View previous topic :: View next topic |
| Do you want systemd as default on Gentoo? |
| I <3 systemd!! I want Gentoo to switch!! |
|
12% |
[ 26 ] |
| Get that horse-crap away from Gentoo as far as possible! |
|
87% |
[ 186 ] |
|
| Total Votes : 212 |
|
| Author |
Message |
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
 Posted: Thu Oct 02, 2014 5:55 am Post subject: |
 |
|
Yeah that mailing list thread is hilarious; the first post reads like a spam email:
| Jóhann B. Guðmundsson wrote: | | I would like you to know that those of us that have been doing systemd distribution integration work (Arch,Fedora,Gentoo,OpenSuse,Mageia etc ) have been collaborating and collectively sharing experience and knowledge among ourselves how the migration process is, has taken place in our distributions as well been collectively sharing units across those distributions, making the transaction to systemd as smooth as possible in the process and I would like you to know that I'm available to assist the Chromium community anyway I can in making the transaction to systemd as smooth as possible should you need it. |
I kept thinking he'd be asking the reader to put money in an account in Nigeria, or something.
It shows the "third-column" nature of the systemd advocates, as well; they really do look like a cult. In fact it reminds me a lot of student-politics, and people going off to get "the party line" so they can come back and carry on arguing. Talk about conviction.. ;)
Love the way they lump Gentoo in there; good to know we've toed the line and made "the transaction to systemd". I must've missed that eselect news. You see the same "argumentum ad populum" as the other post put it, about "all the cool embedded people are switching too", with no mention ofc that they're doing it so their customers can pretend the GPL doesn't exist.
I look forward to that one being tested in a Court. (WTF are their lawyers thinking?! ;-)
| tld wrote: | | I've never fallen for that in my life, yet people seem to be falling for it with sysd. |
Conman^W Wizard's First Rule ;) People are stupid; we believe what we want to believe, even in the face of overwhelming contradictory evidence (cf: evolution/age of the Earth, and man-made global warming.) Cognitive dissonance does most of the work, once you've inculcated enough FUD, since it's the lazy option.
| Quote: | | Besides, if you don't have a lot of unneeded BS, since when is boot time an issue anyway? I'm not sure about others here, but I can boot up to a working fluxbox desktop in about 30 seconds (including web development stuff like apache, MySQL etc) on an 11-year-old x86 dinosaur I'm running here. I'd have to think the same setup on anything current would be like flicking a light switch. I mean talk about a cure for which there's no known disease... |
Hehe indeed; apparently we should cede our desktops and servers, our flexibility and our capability, so that Amazon can save a few pence starting up a VM and dodge even more tax.. Oh wait, openrc is actually better for that, it doesn't constrain your control, and does correct dependency resolution, as well as providing the ability to write tiny, declarative init units; jails, virtualisation, and full cgroup handling, constrained only by the admin. Er, what was this about again? ;p
edit@ almost forgot the one good thing out of that long thread sidetrack; looks like interesting work if quite similar to runscript (which is quite similar to bsd.)
The device services thing might be interesting for the static dev/Olde-Fashioned Gentooee crowd. |
|
| Back to top |
|
 |
happosai
n00b
Joined: 17 Aug 2005
Posts: 43
|
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Thu Oct 02, 2014 7:00 pm Post subject: |
|
|
Yeah Drepper has nothing to do with glibc, and we're better off with Poettering's style of ABI guarantee ("what's an ABI guarantee?") than Drepper's. </sarcasm>
Thanks for sharing. |
|
| Back to top |
|
|
RazielFMX
l33t
Joined: 23 Apr 2005
Posts: 835
Location: NY, USA
|
| Posted: Fri Oct 03, 2014 2:07 pm Post subject: |
|
|
| Shamus397 wrote: | I have to say, that post about SystemD: Biggest Fallacies has some real gold nuggets* in it. And I have to give props to the Google engineers (as much as I despise Google), they're not so easily buffaloed by the SystemD proselytizers. 
I think from now on when anyone posts a 0pointer.de link to me, I'm going to post that 'SystemD fallacies' back.
* phpBB seems to choke when I try to embed the Google Groups URL, so you'll have to cut & paste: https://groups.google.com/a/chromium.org/forum/#!topic/chromium-os-discuss/R4W2fhiwvyg |
The google group thread was extremely interesting. Thank you for sharing. I'm glad there individuals with actual financial clout who see systemd for what it really is (and is not). Maybe RH will be forced to play nice with other init systems after all, which at the end of the day, is all we really want.
Wow. That guy seems like an arrogant d---... |
|
| Back to top |
|
|
happosai
n00b
Joined: 17 Aug 2005
Posts: 43
|
| Posted: Sat Oct 04, 2014 8:44 am Post subject: |
|
|
| RazielFMX wrote: |
Wow. That guy seems like an arrogant d---... |
Drepper's behaviour was the reason why back then an Glibc fork named Eglibc came into existance in 2009 because enough people got fed up enough with his ego, arrogance and therefore staleness in Glibc itself, to which even Debian switched back then. Finally the FSF got fed up enough with Drepper and shunned him in 2012 to replace him with a Glibc steering committee.
Back to topic: I don't see any good reason why Gentoo should eject OpenRC and switch over to Systemd, since one of the fundamentals of Gentoo has always been that we've got a choice. You want Systemd? Fine, emerge it, configure it, run it, have fun with it, you can already have it all the way you want. Just because you want it doesn't translate into "it should be the default init system of Gentoo!"
Systemd in my opinion addressed some valid points, which older init systems didn't care about it and got those right. But the problem with Systemd is also, in my opinion, that they are nowadays way too busy in becoming the new Emacs and still are busy putting stuff into it, where I really do wonder: "Why does this need to be put into an Init system?", meaning it is already a convuluted piece of crap and still continues to walk further down that road.
This is one of the problems about it, it becomes more and more unpredictable, unmaintanable and untrustworthy with every newly added feature IMHO.
The other is that former independant pieces of software liked udev and maybe Networkmanager got sucked up into that unholy piece of crap and maybe unable to run without it in the future (thank god for eudev and wicd!)
Systemd is definitely some piece of software I wouldn't trust my production servers to run with. Making it the default init system out of the box on Gentoo doesn't seem very sane to me. |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6051
Location: Removed by Neddy
|
| Posted: Sat Oct 04, 2014 9:08 am Post subject: |
|
|
Why would Gentoo 'eject' openRC.
Even if Gentoo for some reason pushed systemd as the default init system that's not how Gentoo works.
How many init systems are there in the tree?
How many package managers?
Gentoo takes choice to a whole new level when compared to other more liberal binary distributions
_________________
| Quote: | | Removed by Chiitoo |
|
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sat Oct 04, 2014 9:23 am Post subject: |
|
|
| happosai wrote: | | Systemd in my opinion addressed some valid points, which older init systems didn't care about it and got those right |
I don't agree. Systemd has just put a lot of independent stuff into the init-system which does not belong to an init system.
I understand what you mean: You are speaking about a "daemon-controller". There are several independent daemon-contollers out there, and it is not rocket science to include support for one or several for them into an init system. This is indeed something which has not happened in many init-systems yet.
However, adding such support does not mean that the daemon-controller should be part of the init-system.
For the other parts of system (udev, logging, mounting, network, dhcp, time, ...), it makes no sense at all to be part of the init-systemsm: These are all just poor re-implementations of existing and much better functioning tools, but systemd is used as a political tool to force all people to use that BS.
| Quote: | | But the problem with Systemd is also, in my opinion, that they are nowadays way too busy in becoming the new Emacs |
This is not a good comparison: For an editor, it makes sense to allow a very flexible configuration - hence even a programming language as configuration tool. Emacs was not developed with the goal e.g. to press the existing newsreaders from the market: It really makes sense to have one common configuration/framework for interactive typing-related programs. For systemd, there is no technical reason to include software which does rather orthogonal things. |
|
| Back to top |
|
|
happosai
n00b
Joined: 17 Aug 2005
Posts: 43
|
| Posted: Sat Oct 04, 2014 9:59 am Post subject: |
|
|
| mv wrote: | | happosai wrote: | | Systemd in my opinion addressed some valid points, which older init systems didn't care about it and got those right |
I don't agree. Systemd has just put a lot of independent stuff into the init-system which does not belong to an init system.
I understand what you mean: You are speaking about a "daemon-controller". There are several independent daemon-contollers out there, and it is not rocket science to include support for one or several for them into an init system. This is indeed something which has not happened in many init-systems yet.
However, adding such support does not mean that the daemon-controller should be part of the init-system.
For the other parts of system (udev, logging, mounting, network, dhcp, time, ...), it makes no sense at all to be part of the init-systemsm: These are all just poor re-implementations of existing and much better functioning tools, but systemd is used as a political tool to force all people to use that BS.
|
Well, a daemon controller is something that I for myself would not like to have in an init system, because such a thing almost likely is going to be more worse with less features than the already available ones. I for myself am running Monit as a daemon controller where needed. It is reliable and as fine grained when to and how to restart as I want it to be, it gets the job done, works well enough for me. So no need for me to have it in an init system when I can already have Monit. Otherwise I could take the old trick of running the daemon in a shell loop, which is normally not something wisely to do.
| Quote: | | Quote: | | But the problem with Systemd is also, in my opinion, that they are nowadays way too busy in becoming the new Emacs |
This is not a good comparison: For an editor, it makes sense to allow a very flexible configuration - hence even a programming language as configuration tool. Emacs was not developed with the goal e.g. to press the existing newsreaders from the market: It really makes sense to have one common configuration/framework for interactive typing-related programs. For systemd, there is no technical reason to include software which does rather orthogonal things. |
Well, it is a good comparison if you take into account that Emacs was one of the first well known programs for creeping featurism ("It cannot just cook cofee... yet!") and getting more and more overly complex with all kind of weird stuff. I mean, many people joked about Emacs becoming an operating system some day (somebody btw build such a thing as joke), and for whatever reason Systemd seems to be keen to take that crown today.
Though the thing is Emacs is still a good editor even with that kind of stuff, which you could simply turn off - Systemd seems to be a very different kind of matter to me. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sat Oct 04, 2014 11:01 am Post subject: |
|
|
| happosai wrote: | | Well, a daemon controller is something that I for myself would not like to have in an init system, because such a thing almost likely is going to be more worse with less features than the already available ones. |
Yes, this is what I said: It does not belong to an init-system. It would be nice to have support for it in the init-system nevertheless (i.e. running the daemon-controller instead of the somewhat hackish start-stop-daemon). However, for such a change it is not necessary to develop a new init-system.
| Quote: | | Well, it is a good comparison if you take into account that Emacs was one of the first well known programs for creeping featurism |
I never understand why people speak about "featurism" as if it were something bad: Having features is not a bad thing, because it means you can do things which you could not do before. With systemd it is different: It cannot do anything which was not possible before; it just does the existing things in a nonstandard manner, practically always in a worse way than the existing tools.
| Quote: | | and getting more and more overly complex with all kind of weird stuff. |
It is an enormous difference whether GUI-related things are complex or whether the basic functionality of the system is.
| Quote: | | I mean, many people joked about Emacs becoming an operating system some day |
Yes, but as you say: It is just a joke. Emacs never had plans to do this, and it was certainly never recommended to run it with root permissions so that it is able to override the unix permission systems and run its own on top of the operating system. |
|
| Back to top |
|
|
szatox
Advocate
Joined: 27 Aug 2013
Posts: 3150
|
| Posted: Sat Oct 04, 2014 1:59 pm Post subject: |
|
|
| Quote: | | Having features is not a bad thing |
Yeah, normal people say "if it's not broken, don't fix it.
IT engineers say "if it's not broken, it doesn't have enough features"
Seriously, guys, you're still repeating the same argument over and over and over and over.... and over.. and over again? |
|
| Back to top |
|
|
Dr.Willy
Guru
Joined: 15 Jul 2007
Posts: 547
Location: NRW, Germany
|
| Posted: Sat Oct 04, 2014 2:30 pm Post subject: |
|
|
| mv wrote: | | I never understand why people speak about "featurism" as if it were something bad: Having features is not a bad thing, because it means you can do things which you could not do before. |
People usually say "featurism" when programs go beyond their intended use.
If you can add a feature without violating the rule of modularity, the rule of composition, the rule of simplicity and the rule of parsimony, then by all means add that feature.
If you can't, then it's probably a bad idea to add the feature to this exact program. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sat Oct 04, 2014 2:33 pm Post subject: |
|
|
| szatox wrote: | Yeah, normal people say "if it's not broken, don't fix it.
IT engineers say "if it's not broken, it doesn't have enough features" |
Idleness is the mother of all evil. An evil chance seldom comes alone. Habit cures habit. As it rains in March so it rains in June.
OK, I am convinced. If peoples' wisdom knows that I do not need any features, I am glad to spend pointless hours to solve things which the corresponding feature might have done for me by just specifying an option. |
|
| Back to top |
|
|
happosai
n00b
Joined: 17 Aug 2005
Posts: 43
|
| Posted: Sat Oct 04, 2014 3:39 pm Post subject: |
|
|
| mv wrote: |
I never understand why people speak about "featurism" as if it were something bad: Having features is not a bad thing, because it means you can do things which you could not do before. With systemd it is different: It cannot do anything which was not possible before; it just does the existing things in a nonstandard manner, practically always in a worse way than the existing tools.
|
Because it is bad when some people are adding stuff to a system which was never meant to be there and are trying to reinvent the wheel instead of using the well known and well established tools for that already instead.
To state the jargon file here about creeping featurism:
"[...] the tendency for anything complicated to become even more complicated because people keep saying “Gee, it would be even better if it had this feature too”. (See feature.) The result is usually a patchwork because it grew one ad-hoc step at a time, rather than being planned. Planning is a lot of work, but it's easy to add just one extra little feature to help someone ... and then another ... and another.... When creeping featurism gets out of hand, it's like a cancer. The GNU hello program, intended to illustrate GNU command-line switch and coding conventions, is also a wonderful parody of creeping featurism; the distribution changelog is particularly funny. "
And that's the problem with Systemd: adding up a feature here, there and, oh, over there! But many people running init systems, just like I am, are very conservative about those and therefore watch this with great concerns.
The thing is: what does Systemd really want to be? To quote themselves "systemd is a system and service manager for Linux, compatible with SysV and LSB init scripts."
But it has already left those humble beginnings by a mile and a half at least, together it really more looks like a full blown framework for whatever purpose necesarry. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sat Oct 04, 2014 5:08 pm Post subject: |
|
|
| happosai wrote: | | Because it is bad when some people are adding stuff to a system which was never meant to be there and are trying to reinvent the wheel instead of using the well known and well established tools for that already instead. |
I completely agree that this is bad, and that systemd behaves like that. However, I would not call this featurism: Reinventing the wheel and trying to push well-established tools from the market is a strategy of gaining power and has not much to do with features.
I consider it a completely different thing if an editor/shell/language/... adds some useful function which was not there before: It only improves that tool and does not hurt (except increasing the manpage), and such steps should be encouraged and not vituperated as "featurism" and even be compared with cancer.
For sensitive tools running with root privileges, this is again a different story, of course. |
|
| Back to top |
|
|
Dr.Willy
Guru
Joined: 15 Jul 2007
Posts: 547
Location: NRW, Germany
|
| Posted: Sat Oct 04, 2014 5:37 pm Post subject: |
|
|
| mv wrote: | | I consider it a completely different thing if an editor/shell/language/... adds some useful function which was not there before: It only improves that tool and does not hurt (except increasing the manpage), and such steps should be encouraged and not vituperated as "featurism" and even be compared with cancer. |
That is debatable at best, because simplicity and/or size can be a design value, too.
Take RISC for example. You don't add anything to that instruction set unless you know you have to.
Or any languages standard library. Sure you can add a webserver implementation to libc, but should you? Python has one, C doesn't. It depends on what you want to achieve. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sat Oct 04, 2014 6:12 pm Post subject: |
|
|
| Dr.Willy wrote: | | Take RISC for example. |
This is a very exceptional case and not even software. I agree that memory or speed restrictions might make it necessary to limit some features in certain cases. But this is a pityful limitation. Limiting functionality in such a situation is certainly not a goal but only a method which might be necessary to achieve a goal which might be more important in certain situations than some functionality.
| Quote: | | Or any languages standard library. |
This is a very bad example, since it is more a social question what are "standard" libraries. In the end, it makes no difference if some library is not officially declared as "standard" but can be found on any system in practice, nevertheless. Moreover, for libraries it is much easier to separate things.
In the examples which I gave (editor, shell, language interpreter), if some feature is missing, simply some useful functionality is missing. And this can hardly be desirable. That's why I think especially concerning emacs the word "featurism" is completely wrong: An editor can never have enough features. |
|
| Back to top |
|
|
Dr.Willy
Guru
Joined: 15 Jul 2007
Posts: 547
Location: NRW, Germany
|
| Posted: Sat Oct 04, 2014 7:58 pm Post subject: |
|
|
| mv wrote: | This is a very bad example, since it is more a social question what are "standard" libraries. In the end, it makes no difference if some library is not officially declared as "standard" but can be found on any system in practice, nevertheless. Moreover, for libraries it is much easier to separate things.
In the examples which I gave (editor, shell, language interpreter), if some feature is missing, simply some useful functionality is missing. And this can hardly be desirable. That's why I think especially concerning emacs the word "featurism" is completely wrong: An editor can never have enough features. |
No, this is not a social question. Put standard in quotes as much as you want, a standard is not, or should not be, socially defined.
If you disagree, stop complaining about bash-isms in sh scripts.
As for features in emacs or editors in general: There is the point where it makes more sense to write a new program instead of adding the functionality as a feature to your editor. Unless of course you disagree with modularity in general. |
|
| Back to top |
|
|
ct85711
Veteran
Joined: 27 Sep 2005
Posts: 1791
|
| Posted: Sat Oct 04, 2014 9:01 pm Post subject: |
|
|
The part on programs adding features is fine, if it falls in the part of; does it have a reason for the package's core foundation. It's pretty much, for like emacs to add the perverbial kitchen sink, the perverbial toilet, and perverbial washing machine. Sure, there's going to be tons of people, maybe even the emac's developers too, say it's a new feature. But, what reason does a text editor have for needing a dang kitchen sink!
(Note, I'm just trying to bash emacs, or their developers; I have nothing against them. Only using emacs as an example, since it appears to be a common favorite.) |
|
| Back to top |
|
|
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54308
Location: 56N 3W
|
| Posted: Sat Oct 04, 2014 9:12 pm Post subject: |
|
|
Well, emacs did used to stand for Eight Megs [of RAM] And Constantly Swapping but you will need to be long in the tooth to remember that.
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Sat Oct 04, 2014 9:32 pm Post subject: |
|
|
This, pure and simple:
| Dr.Willy wrote: | If you can add a feature without violating the rule of modularity, the rule of composition, the rule of simplicity and the rule of parsimony, then by all means add that feature.
If you can't, then it's probably a bad idea to add the feature to this exact program. |
As he said, "featurism" means when programs go beyond their intended use. The above gives you a framework to make that decision, and bikeshed it to death.. ;)
Still, that's part of the process of requirements specification; but so is spiking to explore the problem-space. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sun Oct 05, 2014 7:48 am Post subject: |
|
|
| Dr.Willy wrote: | | No, this is not a social question. Put standard in quotes as much as you want, a standard is not, or should not be, socially defined. |
We were speaking about standard libraries, i.e. something where it costs at most one code line to have a "non"-standard library - the only question is whether it is installed. If the social agreement would be to use "boost" everywhere, it would be completely unimportant what the standard C++ comitteee says.
| Quote: | | If you disagree, stop complaining about bash-isms in sh scripts. |
Here the social agreement is to consider POSIX as the standard. I am fine with any features which are POSIX-compatible and which are really features (e.g. <(...)).
In fact, I think bash does not have enough features, that's why I would have preferred to use zsh for ebuilds; or even better perl or python. But we are getting rather OT here.
| Quote: | | As for features in emacs or editors in general: There is the point where it makes more sense to write a new program instead of adding the functionality as a feature to your editor. Unless of course you disagree with modularity in general. |
Besides some obvious jokes, like towers of hanoi, I do not see many examples of this in emacs. You might argue that a mail- or newsreader should be a separate program. Maybe, maybe not: Most mail- and newsreaders bring their own poor editor and perhaps a possibility to call a powerful editor externally for the main text. Both is not very convenient or natural. Thus, if you already have a powerful editor, it really makes sense that this editor optionally can provide the (user-interface for) mail/news-handling. And I can understand that such a user-interface by its nature is so tied to the internals of the editor that it really makes no sense to ship it separately.
I think in these aspects one really must distinguish between system software and a GUI: A GUI by its nature is complex, and any separation is somewhat artificial. But system software must not be for reasons of security and reliability. Again: This simplicity is not a goal per se but a necessary method to reach this higher goal of security/reliability. The bad thing about systemd is that it has sacrified the latter goal completely (and then indeed it does not matter whether feature after feature is added, but one should not mix cause and symptom). |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sun Oct 05, 2014 9:48 am Post subject: |
|
|
| NeddySeagoon wrote: | | Well, emacs did used to stand for Eight Megs [of RAM] And Constantly Swapping but you will need to be long in the tooth to remember that. |
It seems that my teeth are long enough 
And the same interpretation of the phrase nowadays shows that it would have been a wrong decision to cut functionality due to the memory restrictions: There are other editors which you can use if you have a memory restricted system. |
|
| Back to top |
|
|
krinn
Watchman
Joined: 02 May 2003
Posts: 7470
|
| Posted: Sun Oct 05, 2014 11:17 am Post subject: |
|
|
| mv wrote: | | And the same interpretation of the phrase nowadays shows that it would have been a wrong decision to cut functionality due to the memory restrictions: There are other editors which you can use if you have a memory restricted system. |
Well, it makes sense if for you a <=8megs system is a restrict system, but without real limits to put there, i disagree with you.
If we follow your logic, all systems will be memory restrict, because xargs need 16g, cat need 6g and echo will eat its 8g too. And so anyone using the 3 tools would need a little 30g ram or face swapping ; but its ok because cat could browse the web, echo can use encryption and xargs can drives network to use many computers for the task. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6749
|
| Posted: Sun Oct 05, 2014 1:07 pm Post subject: |
|
|
| krinn wrote: | | If we follow your logic ... |
Stay realisitic. Show me examples where features really need gigs of memory. Or at least, since this is clearly impossible, show me an example where features really cause harm. (When looking for examples, keep in mind that "feature" always implies that it is optional and can be switched off.)
I cannot see this in emacs, vim, glibc, bash, zsh, perl, all of which are claimed by some people to have too many features. In fact, all these would loose enormously their value if their feature-richness would be removed. I also cannot see that the features are the problem of systemd - in fact, I am convinced that some people have turned into systemd fanboys because of some features which they particularly like. |
|
| Back to top |
|
|
i4dnf
Apprentice
Joined: 18 Sep 2005
Posts: 271
Location: Bucharest, Romania
|
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Gentoo Chat
