View previous topic :: View next topic |
Author |
Message |
avx Advocate
Joined: 21 Jun 2004 Posts: 2152
|
Posted: Wed Apr 25, 2012 8:46 pm Post subject: |
|
|
Quote: | Open Source happens to follow the path of people with most code/time to spent productively (no, not on forums). | Yep, and that's exactly the problem.
As long as new things/changes are done in a way, the end user can opt-in, it's all dandy, but the way it currently goes, there's no (simple) way to opt-out. The user gets locked into views of others, it's no different than being forced to accept what Microsoft, Apple or Google thinks it's best. Sure, to some degree it's still possible to make the system work, because of the code lying in the open, that doesn't mean that anybody can do it, though.
Most users have an opinion about how there system should work, but they've got limits in their time and abilities to do it themselves and it's getting harder by the day.
IMHO, if these trends continue, there's no reason to choose Linux over OSX/Windows. _________________ ++++++++++[>+++++++>++++++++++>+++>+<<<<-]>++.>+.+++++++..+++.>++.<<+++++++++++++++.>.+++.------.--------.>+.>. |
|
Back to top |
|
|
trippels Tux's lil' helper
Joined: 24 Nov 2010 Posts: 137 Location: Berlin
|
Posted: Thu Apr 26, 2012 3:30 pm Post subject: |
|
|
The next phase has already arrived:
Quote: | We don't support udev in an separate tarball anymore. We do support udev built from the systemd tree used independently of systemd |
|
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Thu Apr 26, 2012 5:54 pm Post subject: |
|
|
MarekSieradzki wrote: | Open Source happens to follow the path of people with most code/time to spent productively (no, not on forums). |
You keep blathering on about this being a meritocracy. I see little merit in breaking working systems for cosmetic whimsy. |
|
Back to top |
|
|
lebel n00b
Joined: 25 Aug 2004 Posts: 36
|
Posted: Fri Apr 27, 2012 1:33 am Post subject: |
|
|
kimmie wrote: | Except... I'm yet to see a convincing argument for having / and /usr as separate partitions, so that's not really a con, it's just an inconvenience. Can you point me a situation where separating / and /usr is of any real advantage? |
It is more of a force of habit and old habits. I've being a sysadmin since '92 and back in those days, when you had a 250MB hard drive you were considered lucky and most of the time, admins were forced to split the partitions in order to fit everything up. So, most essential binaries to boot were put on /bin, /etc, /var and /sbin and /usr was used when you were out of single user and in full multi-user mode. |
|
Back to top |
|
|
broken_chaos Guru
Joined: 18 Jan 2006 Posts: 370 Location: Ontario, Canada
|
Posted: Fri Apr 27, 2012 5:59 pm Post subject: |
|
|
lebel wrote: | So, most essential binaries to boot were put on /bin, /etc, /var and /sbin and /usr was used when you were out of single user and in full multi-user mode. |
This still applies today in some situations. If you have limited local storage (or a very small flash chip for embedded systems) and a big whack of network or hard drive storage, you can fit everything needed on / in about 50 megs even on a full install, which is a heck of a lot harder to do with /usr, /var, and so on. |
|
Back to top |
|
|
MarekSieradzki n00b
Joined: 05 Oct 2010 Posts: 18
|
Posted: Fri Apr 27, 2012 11:00 pm Post subject: |
|
|
Ant P. wrote: | MarekSieradzki wrote: | Open Source happens to follow the path of people with most code/time to spent productively (no, not on forums). |
You keep blathering on about this being a meritocracy. I see little merit in breaking working systems for cosmetic whimsy. |
I didn't use word meritocracy nor did I suggest that decisions only by people that write code with no feedback are the best way to do things. Still, Open Source is usually like I said. |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Sat Apr 28, 2012 7:45 am Post subject: |
|
|
It was probably posted before, but I just found this one to be worth reading (taken from debian-devel mailing list): A few observations about systemd (+ the follow-ups) |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Sat Apr 28, 2012 8:20 am Post subject: |
|
|
Having an init process with many lines of code, a mess of dependencies and a proprietary service specification format is an invitation to disaster.
Boot-up is complicated as it is -- making things complex and rigid is making them fragile, i.e. very non-Unix-y.
If you don't like the fact that much of Unix-like platforms are made of duct tape and zipties, you're on the wrong platform. That is its strength. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Sat Apr 28, 2012 8:57 am Post subject: |
|
|
After reading through the mailing list, I don't think systemd is going to be so bad after all. |
|
Back to top |
|
|
DaggyStyle Watchman
Joined: 22 Mar 2006 Posts: 5909
|
Posted: Sat Apr 28, 2012 11:37 am Post subject: |
|
|
genstorm wrote: | After reading through the mailing list, I don't think systemd is going to be so bad after all. |
afaik systemd isn't the problem, the fact that we need to dedicate part of the memory for that crappy thing called initramfs if we want to properly boot with newer udev. _________________ Only two things are infinite, the universe and human stupidity and I'm not sure about the former - Albert Einstein |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Sat Apr 28, 2012 11:43 am Post subject: |
|
|
genstorm wrote: | After reading through the mailing list, I don't think systemd is going to be so bad after all. |
If systemd didn't have the problems pointed out by the person on the mailing list, which I repeat above, it might have been worth taking seriously.
This *kit/dbus mess is not a reliable foundation for a solid operating system. The init daemon should be a tiny process with zero bugs and dependencies, which then bootstraps the rest of your boot system.
Perhaps someone could explain to me why people feel compelled to make these compromises in sound design to save a few seconds on boot, which is something I do once every several months on my laptop. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Sat Apr 28, 2012 11:43 am Post subject: |
|
|
@DaggyStyle: That's the immediate trouble we have with udev, yes, but especially since the merge of udev into the systemd tarball this is also about the latter. Personally I will move my /usr into rootfs when there's time for that, it was a 'good design' decision back then but isn't really needed on my notebook system.
Hypnos wrote: | genstorm wrote: | After reading through the mailing list, I don't think systemd is going to be so bad after all. |
If systemd didn't have the problems pointed out by the person on the mailing list, which I repeat above, it might have been worth taking seriously. |
Read further, not just the first post. |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Sat Apr 28, 2012 1:13 pm Post subject: |
|
|
genstorm wrote: | Read further, not just the first post. |
I did, and the problems revealed get worse: not just Linux-specific, but requiring a minimum kernel version, and having "features" such as inetd emulation.
You can read Poettering's own manifesto on systemd here. In there is a list of features which he thinks are good things.
Again, how is this consistent with the Unix philosophy of having small tools that each do one thing well, and then tying them together to do what *you* want? For what benefit, saving a few seconds every 6 months when you upgrade your kernel?
Just like with PulseAudio, he offers a solution to a problem we don't have and after having to adopt it makes our systems more complex and fragile. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Sat Apr 28, 2012 1:32 pm Post subject: |
|
|
udev has required minimum kernel versions in the past already. udev's current (Gentoo arch is at udev-171-r5) kernel minimum is 2.6.32, the latest systemd package in portage requires 2.6.38, current longterm stable kernel is 3.0, so what? systems that keep outdated kernels around may not need to be the first to migrate to systemd as well.
As for your other criticism, also taken from the mailing list:
Quote: | it uses modprobe, it uses mount and setting the host name has always
been part of the boot process, so that init is doing this is just a
change in the sense that it's done by C code rather than a shell script,
so in those cases it just does what is already done by a bunch of shell
scripts.
as for trying to take over inetd's role: yes, it can replace inetd. You
don't have to do that, though. Replace cron and at? Not really, it has
support for timer events, but it does not have any replacement for the
user part of at and cron, so while it can do some of what at and cron
can, it's currently no replacement for them. |
It seems like getting rid of dozens of separate shell scripts while still retaining the possibility to use them is a neat idea. Plus, according to the opening post, systemd is very well documented (which kind of weakens the point of Red Hat fearmongers). |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Sat Apr 28, 2012 2:31 pm Post subject: |
|
|
genstorm wrote: | udev has required minimum kernel versions in the past already. udev's current (Gentoo arch is at udev-171-r5) kernel minimum is 2.6.32, the latest systemd package in portage requires 2.6.38, current longterm stable kernel is 3.0, so what? systems that keep outdated kernels around may not need to be the first to migrate to systemd as well. |
More relevant is that udev-114 is supported, later versions can be built without bloat like gobject introspection, and there are drop-in replacements if you don't need hotplugging -- still Unix-y at the end of the day.
Quote: | It seems like getting rid of dozens of separate shell scripts while still retaining the possibility to use them is a neat idea. |
No, it's called bloat. The point of having separate shell scripts is that they can be separately maintained and tweaked without having to send patches upstream.
Modularity == choice == good (== Gentoo)
Quote: | Plus, according to the opening post, systemd is very well documented (which kind of weakens the point of Red Hat fearmongers). |
That is good, but it's more important that the wider community be involved in its code development maintenance, esp. because it's does so many things. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
MarekSieradzki n00b
Joined: 05 Oct 2010 Posts: 18
|
Posted: Sun Apr 29, 2012 12:08 am Post subject: |
|
|
Hypnos wrote: | Having an init process with many lines of code, a mess of dependencies and a proprietary service specification format is an invitation to disaster.
Boot-up is complicated as it is -- making things complex and rigid is making them fragile, i.e. very non-Unix-y.
If you don't like the fact that much of Unix-like platforms are made of duct tape and zipties, you're on the wrong platform. That is its strength. |
Not at all, Unix is horrible but it looks nice when compared with absolutely ridiculous things like Windows. Shell scripts are horrible (as a language) and slow. I guess DSL hype hasn't come to Unix admins yet.
Quote: |
"I liken starting one's computing career with Unix, say as an undergraduate, to being born in East Africa. It is intolerably hot, your body is covered with lice and flies, you are malnourished and you suffer from numerous curable diseases. But, as far as young East Africans can tell, this is simply the natural condition and they live within it. By the time they find out differently, it is too late. They already think that the writing of shell scripts is a natural act."
Ken Pier, Xerox PARC |
from http://www.art.net/~hopkins/Don/unix-haters/handbook.html
I like and use Gentoo because of specific advantages of doing so and not braindead following of things that happened by coincidence. Discussion with axiom "Unix way (old way) is right" is as stupid as one with "Poettering is right". |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Sun Apr 29, 2012 2:00 am Post subject: |
|
|
MarekSieradzki wrote: | Ant P. wrote: | MarekSieradzki wrote: | Open Source happens to follow the path of people with most code/time to spent productively (no, not on forums). |
You keep blathering on about this being a meritocracy. I see little merit in breaking working systems for cosmetic whimsy. |
I didn't use word meritocracy nor did I suggest that decisions only by people that write code with no feedback are the best way to do things. Still, Open Source is usually like I said. |
You're nitpicking the definition of words used by me and others while ignoring the actual point of those you're replying to and posting in a very short and concentrated timespan almost solely to defend systemd. I question the motivation behind this. |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Sun Apr 29, 2012 7:35 am Post subject: |
|
|
MarekSieradzki,
Of course reasonable people can disagree about OS design philosophies, and it's a debate worth having. But it's annoying when, having chosen a platform in large part because of its design philosophy, it changes under my feet. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
mv Watchman
Joined: 20 Apr 2005 Posts: 6747
|
Posted: Sun Apr 29, 2012 8:26 am Post subject: |
|
|
Hypnos wrote: | But it's annoying when, having chosen a platform in large part because of its design philosophy, it changes under my feet. |
Yes, that's exactly the point. Consider e.g. the permission concept of Unix. It certainly has its weaknesses and limitations, and so it is not surprising that there are various attempts (SELInux, grsecurity, POSIX capabilities, to name a few involving different issues) to change it. But these projects all change it on the correct place, namely in the kernel and introducing the changed design decisions from ground up.
In contrast, the *kit and systemd "philosophy" is just a plain misuse of the powers of root: Running full-permission complex daemons with the ability to do almost everything is what you should do in no circumstances according to Unix design, because you just add an additional attack vector (namely all possible bugs of that new system). Roughly speaking: To all the weaknesses of the previous system you just add the weaknesses of the new system.
This is obviously so plain stupid that experienced programmers can just shake their head how you might even consider for more than a second running such a thing. |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Sun Apr 29, 2012 9:24 am Post subject: |
|
|
mv wrote: | [...] But these projects all change it on the correct place, namely in the kernel and introducing the changed design decisions from ground up. [...] |
Well said. This is something the BSD community does well. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
depontius Advocate
Joined: 05 May 2004 Posts: 3509
|
Posted: Sun Apr 29, 2012 8:03 pm Post subject: |
|
|
It may please people here to read this:
http://lwn.net/Articles/492125/
Some aspects of what Lennart is trying to do got smacked rather hard by Linux. Other parts he liked. IMHO, the resulting direction looks pretty decent.
I'll say again, there are 3 responses to a lot of what is happening here:
1 - Go along with it, without thinking too hard.
2 - Fight it and refuse to go along with it.
3 - Do what can be done to add sanity to it, hopefully an adequate amount of sanity.
Most of what I have seen here and in this thread ( https://forums.gentoo.org/viewtopic-t-921140-postdays-0-postorder-asc-start-50.html ) have been #2. While I don't inherently think that's a bad idea, unless people band together and come up with a defined project implementation, it will just be a bunch of refuseniks. Put together as a project, it has real possibilites.
I'm trying to look in to #3, which could start with simply better mid-level documentation, continuing to debug utilities, both gui and text mode. So far this idea has gotten no traction with others. I think of it as the "Development in Vermont" problem. There are 3 camps here - The first came wants to plan for development, and do it properly. The second camp wants no development, and spends its time fighting the first camp. The third camp just wants to make money, and goes off an develops while the first 2 camps are fighting each other. As a result, much of the development is unplanned and haphazard. _________________ .sigs waste space and bandwidth |
|
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
|
|