Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Systemd headsup of L.P. - reads like at first of april
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, 4, 5, 6, 7  Next  
This topic is locked: you cannot edit posts or make replies.    Gentoo Forums Forum Index Gentoo Chat
View previous topic :: View next topic  
Author Message
ulenrich
Veteran
Veteran


Joined: 10 Oct 2010
Posts: 1159

PostPosted: Mon Oct 07, 2013 10:10 pm    Post subject: Reply with quote

grey_dot wrote:
ssuominen wrote:

That's quite the opposite what would help for embedded; what would help is completed /usr merge so bin, sbin, lib, lib32, lib64, and other package-manager only-maintained directories would all be in the same partition for easy read-only mounting, like for embedded and sdcard.
As in, all this is to *help* embedded, not to take anything away from it.

You are wrong.

No, you are wrong!
And you are wrong in thinking your say without any reasonable argument has any value at all.
_________________
fun2gen2
Back to top
View user's profile Send private message
dmpogo
Advocate
Advocate


Joined: 02 Sep 2004
Posts: 2289
Location: Canada

PostPosted: Tue Oct 08, 2013 5:48 am    Post subject: Reply with quote

grey_dot wrote:
dmpogo wrote:
Exactly, desktop is one area where Linux is mostly irrelevant but somehow it manages to dictate how one should boot one's computer :)


Dictate who? Me? No. Millions of windows users? Certainely no. Linux does not dictate anything, just a bunch of people fu^Wscrews everything up as usual.


Hm, I meant desktop dictates Linux development
Back to top
View user's profile Send private message
ssuominen
Developer
Developer


Joined: 30 Sep 2005
Posts: 2132
Location: Finland

PostPosted: Tue Oct 08, 2013 7:19 am    Post subject: Reply with quote

saellaven wrote:
Just because your system is working today doesn't mean it will work tomorrow because Gentoo has decided to follow upstream even if it means going off a cliff. And those that do strive to achieve the previous behaviors will be marked with outright hostility and derision, as ssuominen has routinely done to eudev and now despite breaking his implied promise that a separate /usr would stay when he stabilized >udev-200ish (while further deriding eudev in the same message).


I've told this before in this thread, and I'm telling it again -- separate /usr has absolutely nothing to do with sys-fs/udev and sys-fs/udev is installed to / and will continue to be installed to /
So exactly what promise I have broken? None. Stop your foundless accusations.
Back to top
View user's profile Send private message
steveL
Advocate
Advocate


Joined: 13 Sep 2006
Posts: 2709
Location: The Peanut Gallery

PostPosted: Tue Oct 08, 2013 9:14 am    Post subject: Reply with quote

ssuominen wrote:
saellaven wrote:
those that do strive to achieve the previous behaviors will be marked with outright hostility and derision, as ssuominen has routinely done to eudev

Stop your foundless accusations.

Without getting into what you have or have not promised, the above description of your insults to the eudev team is entirely accurate, and reflects what I described elsewhere as your agenda. And it's such a shame, since they would have loved to work with you when they started out, and you could have nurtured them as a Gentoo sub-project which you merely advised, and had no responsibility for.

Very badly-handled, as a collective: cack-handed, in fact.
Back to top
View user's profile Send private message
ssuominen
Developer
Developer


Joined: 30 Sep 2005
Posts: 2132
Location: Finland

PostPosted: Tue Oct 08, 2013 10:32 am    Post subject: Reply with quote

steveL wrote:
ssuominen wrote:
saellaven wrote:
those that do strive to achieve the previous behaviors will be marked with outright hostility and derision, as ssuominen has routinely done to eudev

Stop your foundless accusations.

Without getting into what you have or have not promised, the above description of your insults to the eudev team is entirely accurate, and reflects what I described elsewhere as your agenda. And it's such a shame, since they would have loved to work with you when they started out, and you could have nurtured them as a Gentoo sub-project which you merely advised, and had no responsibility for.

Very badly-handled, as a collective: cack-handed, in fact.


You must have read wrong, I haven't at any time insulted the eudev team -- I was merely pointing out the facts about it's codebase. Unless you are implying facts are insults?
Back to top
View user's profile Send private message
grey_dot
Tux's lil' helper
Tux's lil' helper


Joined: 15 Jul 2012
Posts: 142

PostPosted: Tue Oct 08, 2013 3:29 pm    Post subject: Reply with quote

ulenrich wrote:
No, you are wrong!
And you are wrong in thinking your say without any reasonable argument has any value at all.


Speaking of value - could you point out the value of your post? Because I don't see any reasonable argument from you and I find the value of your forum registration very questionable, not mentioning the value of your very existence which you may cease without affecting the outcome of this thread (btw does it have any value?).
Back to top
View user's profile Send private message
ssuominen
Developer
Developer


Joined: 30 Sep 2005
Posts: 2132
Location: Finland

PostPosted: Tue Oct 08, 2013 4:30 pm    Post subject: Reply with quote

grey_dot wrote:
ssuominen wrote:

That's quite the opposite what would help for embedded; what would help is completed /usr merge so bin, sbin, lib, lib32, lib64, and other package-manager only-maintained directories would all be in the same partition for easy read-only mounting, like for embedded and sdcard.
As in, all this is to *help* embedded, not to take anything away from it.


You are wrong.


http://fedoraproject.org/wiki/Features/UsrMove

Search the page for word 'read-only' multiple times and you'll find the relevant parts which explains the benefits of having everything in /usr to embedded (and more).
Back to top
View user's profile Send private message
TomWij
Developer
Developer


Joined: 04 Jul 2012
Posts: 1551

PostPosted: Tue Oct 08, 2013 5:45 pm    Post subject: Reply with quote

saellaven wrote:
TomWij wrote:

Exactly, as we can't promise it; we clearly state right away that we can't, because some of those things are broken upstream. That's why we opt for a stable subset instead.


Then you might want to tell your fellow devs to stop making promises as such...


Which promises?

saellaven wrote:
Quote:
If you don't want to follow that stability; then you are on your own, and that is much more compromising than following a system that is widely supported and everyone runs without much problems.


And when systems go unstable because following upstream was easier than doing the right thing, you and the users will get to deal with the headaches caused by it.


Yes, see the recent OpenRC netifrc split causing bonding breakage as an example.

saellaven wrote:
Quote:
Please read the news message again, you can have a separate /usr using an initramfs; and why does that merge even matter?


I've stated my reasons why. Those reasons don't apply to you, so you don't care. I could spend the next week arguing with you and it won't matter. Others could spend the next week arguing with you and it won't matter. The devs have spoken via the council and that's all that matters.


No, you didn't, you always left out initramfs; which makes it a different case you haven't talked about. Please don't assume I don't or won't run an initramfs for a separate /usr.

saellaven wrote:
Quote:
We do add local init files because they are a small addition (and not a patchset); so, if lots of upstream drop them I don't think it is going to be a problem.


So, on one hand, you claim you can't maintain all 18,000 packages and then turn around and say you can handle all 18,000 packages without a problem. How about a little consistency?


You have simplified the consistency away; the difference is still there in my quote, "small addition (and not a patchset)". There's a huge difference there; not in size alone, but also in being able to upstream the effort.

saellaven wrote:
Quote:
Because the boot is just a small picture of the full picture, please don't ignore the rest; also, there is much more to worry about (stability, security, proper testing, quality assurance, prefix, dependency tree, ...).


and when we get to the point where "well, every other distro uses systemd, so we're going to deprecate using openrc, sysvinit, etc because everyone else has abandoned it and we can't test it alone..." ?


When nobody can push them forward anymore?

saellaven wrote:
Quote:

saellaven wrote:
Quote:
Quote:
Quote:

Well, it was about predictable network names, unless I have misunderstood; anyhow, I don't run an initramfs at all, so I indeed don't know what that is all about. Sounds like either users not following news and/or post installation messages or maybe the absence of such messages; or people running ~arch (which isn't covered by such messages until it hits people).


and it utterly fails at that job, as has been documented extensively here on the forum...


Support or documentation doesn't mean it fails to do its job, there is much more of that about the previous network names on the internet.


Several people here have utterly destroyed the entire argument behind it and have shown how it falls flat on its face at its intended purpose... but again, let's follow upstream even when they walk off a cliff and intentionally break systems in unpredictable ways (because adding a peripheral card or changing my USB port should totally change my network device's name) because being a lemming and following upstream is less work.


Not sure which network names you are talking about, so I'll assume the previous network names; if you then still break your system by changing hardware, you haven't prepared and configured it properly.

Gentoo Linux is exactly not the distribution where you can just insert or change a device or card and expect everything to magically just work; it has never been this way, why expect it to be different for network cards?

But that's not entirely true as predictable network names do a good job at keeping it working if you keep the card in its place.


No... and that's where you're wrong. The NEW network naming is neither predictable NOR persistent. Change one card in your system and it can reorder your pci bus, surprisingly changing your network device name. Likewise, if you're using a USB wifi adapter and plug it into a different port, you will, again, get a different name.

We could have defaulted to naming by MAC address (which is supposed to be a unique key), but instead it was chosen to name things by device slots. I install a new network card, I know I'm going to have to change my network configuration. I add a new device to my system, I'm not necessarily expecting my networking to break.

and that's the type of shortsightedness Poettering and friends give us and that's a big part of why I don't want them anywhere near my system. Yet, Gentoo is going down the path of selling out to them. The initramfs "solution" is just as prone to unexpected breakage as the "new and improved" network naming is.


Mine doesn't reorder, location means location and if you thus change one card then the location of the other card won't change; who knows your hardware is problematic. How can things be predictable or persistent when your hardware reports different values? My hardware reports the same properly, and it results in predictable and persistent behavior for me; so the problem certainly is not with the new network names.

The new network names allow you to choose between MAC address and device slot location. If you want to have per actual card settings, you could opt for the former. If you want to swap a network card by one of the same type, you could opt for the latter. But if both of these don't show to be consistent in your hardware; then please don't blame the new network names for it, because they reflect the way your hardware works.

saellaven wrote:
Quote:
saellaven wrote:
So when lvm got updated and it didn't tell you that it would break your existing initramfs unless it's rebuilt, it's the fault of the user... how often until we're rebuilding the initramfs once or twice a week because every package that could potentially be in it throws a warning by default?


If it didn't, please report that; because it is always the intention to do as such. Rebuilding it once or twice a week is normal on a rolling release distribution.


You know when the last time I had my system break because I didn't rebuild my initramfs? Never... as I'm sure is the case for the majority of Gentoo users. So, you, up front, admit that forcing users onto an initramfs is a headache that will potentially unexpected break their systems, but continue defending it as a good thing. I'd call a useless, new point of failure a bad thing, but that's just me.


How sure are you? Do you have data and statistics?

Breaks are not unexpected, which has been clarified before; I am not going to repeat in detail that users are instructed when needed...

It is indeed a good thing for you if you want a separate /usr, I'm not convinced as to why it is not; it is not so much for me, because I don't have interest in the separate /usr configuration yet.

As initramfs works, it allows me to support a lovely /usr configuration if I want or need one; it would be useful to me and not for just me alone.

saellaven wrote:
Quote:

The stone age where we wrote on the walls is over; please consider to upgrade your system properly, instead of praying that running a upgrade command on your exotic configuration will not break things on reboot.

My system works, because I do as such; there isn't anything hard about that. Please realize that we live in a world that changes; what you run today is incompatible with what was available when you were born.

Separate /usr is still here (with initramfs), I don't see the problem. Promises that single developers make about single packages, do not necessarily reflect Gentoo; also note that individual support hasn't been disallowed...


and again, the dismissive arrogance... and you wonder why some people don't want to be Gentoo devs. How long before Gentoo goes the way of GNOME and removes anything that people might find useful because it's just too complex to leave things that have worked working?


We can't guarantee stability for users that don't upgrade and don't use their system properly; this isn't about us being dismissive, this is rather them accepting matters. If those users ditch away our informative and admissive news and messages, go for dismissive odd system configurations that nobody else follows and is known to be unstable, then what more can we do? Not sure what this has to do with becoming a Gentoo Developer.

Is there any indication that we are removing useful or too complex matters (or are going to)?

The only case where that really happens is with Gentoo Security; and even then, we keep things around masked. Other occasions are where versions have been succeeded by other versions; because really, I think we can be very certain that not even 1% still wants to run years old Linux kernels (<2.4). And even then, we help with forward porting where possible; but then again it depends, how much we can pull off. Not everything is possible given limited resourecs; so yes, some things are simply too complex to keep maintaining for a single user that doesn't want to follow the thousand other users.

You can't build a great distribution without making sacrifices, that is impossible; so, a take home question if you want to free up our time: What do you think we are currently providing for certain users that we shouldn't and why?
Back to top
View user's profile Send private message
TomWij
Developer
Developer


Joined: 04 Jul 2012
Posts: 1551

PostPosted: Tue Oct 08, 2013 6:04 pm    Post subject: Reply with quote

grey_dot wrote:
ulenrich wrote:
grey_dot wrote:
You are wrong.
No, you are wrong!
And you are wrong in thinking your say without any reasonable argument has any value at all.


Speaking of value - could you point out the value of your post? Because I don't see any reasonable argument from you and I find the value of your forum registration very questionable, not mentioning the value of your very existence which you may cease without affecting the outcome of this thread (btw does it have any value?).


Come on guys, stating "You are wrong." is a contradiction which is a disrespectful disagreement because your opposing case does not provide any supporting evidence and thus your opposing case has no weight; if you don't have any value to contribute to that weight, please do not respond as this is everything but constructive for the discussion that is being held.
Back to top
View user's profile Send private message
desultory
Administrator
Administrator


Joined: 04 Nov 2005
Posts: 8005

PostPosted: Wed Oct 09, 2013 4:16 am    Post subject: Reply with quote

grey_dot wrote:
ulenrich wrote:
No, you are wrong!
And you are wrong in thinking your say without any reasonable argument has any value at all.


Speaking of value - could you point out the value of your post? Because I don't see any reasonable argument from you and I find the value of your forum registration very questionable, not mentioning the value of your very existence which you may cease without affecting the outcome of this thread (btw does it have any value?).
Dial back the rhetoric, this is your only warning.
Back to top
View user's profile Send private message
grey_dot
Tux's lil' helper
Tux's lil' helper


Joined: 15 Jul 2012
Posts: 142

PostPosted: Wed Oct 09, 2013 11:26 am    Post subject: Reply with quote

ssuominen wrote:
grey_dot wrote:
ssuominen wrote:

That's quite the opposite what would help for embedded; what would help is completed /usr merge so bin, sbin, lib, lib32, lib64, and other package-manager only-maintained directories would all be in the same partition for easy read-only mounting, like for embedded and sdcard.
As in, all this is to *help* embedded, not to take anything away from it.


You are wrong.


http://fedoraproject.org/wiki/Features/UsrMove

Search the page for word 'read-only' multiple times and you'll find the relevant parts which explains the benefits of having everything in /usr to embedded (and more).


I'm sure Fedora is widely used as an embedded linux distribution. In some distant galaxy I don't probably want to know about. Meanwhile, overall experience shows that moving non-essential software to a separate partition might be very useful because of the amount of space it consumes - my n900 (does it count as an embedded?) ubi / partition is only 200mb and contains only softwares essential to boot and some other stuff I might need, and more than half of it is still free. And no, you generally are not advised to move core stuff like kernel modules, often used core utils, etc to emmc/sd card, because those are goddamn slow.

Btw, in a real embedded world (not your average android phone) there generally are only a kernel without modules and a bunch of statically built binaries with maybe some basic glue shell scripts, so standard hier is not applicable in this case.

P.S. The page you linked does not contain the word 'embedded'. Any other relevant arguments you wanna share?
Back to top
View user's profile Send private message
grey_dot
Tux's lil' helper
Tux's lil' helper


Joined: 15 Jul 2012
Posts: 142

PostPosted: Wed Oct 09, 2013 11:56 am    Post subject: Reply with quote

desultory wrote:
grey_dot wrote:
ulenrich wrote:
No, you are wrong!
And you are wrong in thinking your say without any reasonable argument has any value at all.


Speaking of value - could you point out the value of your post? Because I don't see any reasonable argument from you and I find the value of your forum registration very questionable, not mentioning the value of your very existence which you may cease without affecting the outcome of this thread (btw does it have any value?).
Dial back the rhetoric, this is your only warning.


Why? I'm bored.
Back to top
View user's profile Send private message
Yamakuzure
Veteran
Veteran


Joined: 21 Jun 2006
Posts: 1402
Location: Bardowick, Germany

PostPosted: Thu Oct 10, 2013 3:19 pm    Post subject: Reply with quote

steveL wrote:
Basically anyone who wants to tell me how to setup my machine, or to ignore my experience can STFU. I ain't interested one iota.
Actually that pretty much summed up your argument. I quoted from a site linked in a news article available using "eselect news". So no, I am not wrong, because I am not the author of the article. I quoted your statement you *knew* that Poettering and Sievers are behind the /usr merge per se, and that's a false claim. If I misunderstood you, I am sorry. If I did not misunderstand you, please cool down and re-read.
_________________
systemd - The biggest fallacies
Back to top
View user's profile Send private message
steveL
Advocate
Advocate


Joined: 13 Sep 2006
Posts: 2709
Location: The Peanut Gallery

PostPosted: Fri Oct 11, 2013 4:57 am    Post subject: Reply with quote

Yamakuzure wrote:
Actually that pretty much summed up your argument. I quoted from a site linked in a news article available using "eselect news". So no, I am not wrong, because I am not the author of the article. I quoted your statement you *knew* that Poettering and Sievers are behind the /usr merge per se, and that's a false claim. If I misunderstood you, I am sorry. If I did not misunderstand you, please cool down and re-read.

Actually no, you misinterpreted what I said, assuming that it was not in Linux context. I stand by what I said: in the Linux ecosystem the /usr merge has been pushed by Poettering, Sievers et al under the auspices of freedesktop. That's why all the so-called "reference" links about it are their turpid justifications.

Feel free to continue focussing on the strawman of whether Solaris made that move several years previously: it had nothing to do with what I said, nor will it ever; like I said, I don't see that as at all relevant: or we might as well imitate Windows, except Poettering et al have already done that for you.

It may have been relevant in the original argument, but only to examine why Solaris did it, and whether the same considerations apply to the same extent on Linux. I did that years ago when the argument was presented, and it was just as crap then as it is now. So no, that has nothing to do with what I was discussing. I'm sorry you have such a hard time simply accepting my word, but afaic that's your problem, not mine.

Please cool down and re-read what I wrote, since you're so insistent on putting meaning into my words that was not actually there, and indeed refuse to accept clarification when it's put to the question. IMO that's just trolling.
Back to top
View user's profile Send private message
steveL
Advocate
Advocate


Joined: 13 Sep 2006
Posts: 2709
Location: The Peanut Gallery

PostPosted: Fri Oct 11, 2013 5:04 am    Post subject: Reply with quote

ssuominen wrote:
You must have read wrong, I haven't at any time insulted the eudev team -- I was merely pointing out the facts about it's codebase. Unless you are implying facts are insults?

Not at all: stating that you don't trust the people doing the fork (amongst other FUD that I am not alone in noticing) is an insult. Especially when they're Gentoo devs.

Again, cack-handed at best.
Back to top
View user's profile Send private message
_______0
Guru
Guru


Joined: 15 Oct 2012
Posts: 521

PostPosted: Fri Oct 25, 2013 11:38 am    Post subject: Breaking news!! Reply with quote

In a leaked irc conversation is rumored that Lennard Potring is about to give an earth shattering announcemet.

Lennard Potring with the help of his apparatchik has been hard at work to complete his ultimate vision. Irked how things are now with linux Potring will announce:

SystemdOS

Yes, you heard this right. Lennard, tired with the "everything is a file" 'nix philosophy, will absorb practically every nook and crany of an OS into his systemd project.

Details are scant but according to the leaked info traditional 'nix commands will be integrated into systemd.

Things like awk, |, ls, will be all part of systemctl command like this:

systemctl-awk
systemctl-|
systemctl-ls

/etc folder will dissapear as all config file will be directly coded into C within systemd. Potring argues that this way Linux gains in performance, reliability and just about any positive adjective you can think off. Potring promises that no bug will affect the whole. The argument for such an /etc integration is that anything broken package will immediately be noticed as it will affect the whole. This is a technique to notice age old bugs in /etc and thus fixing it for systemd.

Linux package managers will no longer be necessary as systemd will take care of install un-install. For legacy people (not modern and nostalgic) systemd will keep obsolete aliases such as:

systemctl-zypper
systemctl-apt-get
systemctl-pacman
systemctl-emerge
etc...

In the info leaked there something about filesystem too. Potring sees way too many filesystem and his SystemOS does away with all of them and in place will be systemctl-fs.

This amazing announcemt will prove to be very controversial but he argues that by integrating everything he can eliminate wrong ways of doing things and eventually old farts will learn to love SystemdOS.

But Lennart ain't stopping here with SystemdOS. In a separate private irc conversation it is said that the integration of the INTERNET into systemd is already in the works.

Stay tuned!
Back to top
View user's profile Send private message
ulenrich
Veteran
Veteran


Joined: 10 Oct 2010
Posts: 1159

PostPosted: Fri Oct 25, 2013 12:19 pm    Post subject: Reply with quote

Systemd trend is:
- much as possible shouldn't be coded at all but declared
- every line of such declaration should fit into its own file.
- provide enough commodity to easily create new distributions

If the guy named Peottering shows up with a new full blown distribution,
it would be a show case demonstrating exactly that. But I doubt it, because I see him focusing actual problems of Systemd and comitting every day to just that (and not more). And every issue beyond todays focus is added to TODO list of Systemd, which is why that file constantly grows.

Please discuss issues of "SystemOS" following a path contrary to Systemd in another thread.
_________________
fun2gen2
Back to top
View user's profile Send private message
steveL
Advocate
Advocate


Joined: 13 Sep 2006
Posts: 2709
Location: The Peanut Gallery

PostPosted: Fri Oct 25, 2013 9:39 pm    Post subject: Re: Breaking news!! Reply with quote

_______0 wrote:
Stay tuned!

Lol, brilliant.
Back to top
View user's profile Send private message
wuzzerd
Guru
Guru


Joined: 05 Jan 2005
Posts: 433
Location: New Mexico

PostPosted: Sat Oct 26, 2013 12:01 am    Post subject: Re: Breaking news!! Reply with quote

I would believe _______0 if the proposed commands were longer. After two weeks of playing with systemd in a hermetically sealed partition things should read like:

systemctl.awk.oldunix.target operateupon mount-mnt-usr-local-src-myfile systemctl.standardoutput.redirect....

In my senility I forget what I was doing with such commands.
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2370
Location: Dallas area

PostPosted: Sat Oct 26, 2013 12:51 am    Post subject: Reply with quote

Too funny.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.15.9-zen, glibc-2.17, gcc-4.7.3-r1, eudev
xorg-server-1.16, openbox w/lxpanel, nouveau, oss4
Back to top
View user's profile Send private message
mv
Advocate
Advocate


Joined: 20 Apr 2005
Posts: 4219

PostPosted: Sat Oct 26, 2013 9:57 am    Post subject: Reply with quote

ulenrich wrote:
Systemd trend is:
- much as possible shouldn't be coded at all but declared

Which means that as much as possible is hardcoded, and you re left only to select a few fixed choices through your declarations.
Quote:
- every line of such declaration should fit into its own file

Yes. This means in particular that you cannot write a "universal" configuration for all your systems which can be easily changed by "meta"-variables (which the user determines) but that you have to change dozens of files manually (and not only chang but also add or remove files) when you want a different mode. For example, openrc's /etc/conf.d/modules could be identical on all systems, if one uses some shell tests on existing modules, software, hardware, and such "meta"-variables (e.g. in terms of "magic" files) to determine which actually should by loaded. With systemd one must hardcode for every machine and for every purpose a fixed (unconditional) list. Even worse, if e.g. you want to change the laptop from a "home" setup to a "dhcp" or "wireless" or "institute's" or "new hotel" setup - which requires normally not only network changes but also different rules for forwarding, iptables, etc, maybe also decide whether networkmanager or a fixed setup is used, etc: With openrc one could use e.g. "magic" files as a metavariable to script which config's to use, with systemd dozens of configs have to be changed, since only fixed declarations are possible. All improvements of comfort in recent years have been wiped by systemd and we are back to middle ages methods, leaving the user only the choice to press some buttons or not...
Yep, that's the systemd philosophy: Invent everything from scratch, dumping 30 years of experience and permenant improving. Technically, it is pushing us about 30 years back. But yes, it must be much better, because it is newer. What do all these idiots with 30 years experience know anyway?
Back to top
View user's profile Send private message
TomWij
Developer
Developer


Joined: 04 Jul 2012
Posts: 1551

PostPosted: Sat Oct 26, 2013 10:03 am    Post subject: Reply with quote

Well, since NetworkManager can do that, systemd doesn't have to do that and therefore you don't need what you describe; because systemd is just a system daemon, not a network manager.
Back to top
View user's profile Send private message
mv
Advocate
Advocate


Joined: 20 Apr 2005
Posts: 4219

PostPosted: Sat Oct 26, 2013 10:19 am    Post subject: Reply with quote

TomWij wrote:
Well, since NetworkManager can do that

Network manager - apart from being not usable on systems with a reasonable security policy due to polkit dependency, and apart from not being even able to do a simple setup like a fixed local address + a dhcp-controlled outgoing address on the same ethernet-card - is not meant to set forwarding rules, fine-grained iptables rules, mount encrypted cerain partitions which are meant only for the home network, loading appropriate encryption modules, etc pp.
It should be the purpose of the init-system to decide (based on tests determined by the user) whether networkmanager or something else is used and how that is used.
systemd just fails in very basic tasks of an init-system. The only thing you can reasonable do with it is to run your own script which does the whole initialization, which means, if you want to have a reasonable customization, you practically have to write the complete init-system on your own and mask most of the tasks which systemd does faulty on its own.
But then, wouldn't it be more reasonable to use a modern and configurable init-system like openrc from the very beginning instead of the systemd BS with its retarded *.ini-filles as only configuration possibility?
Back to top
View user's profile Send private message
TomWij
Developer
Developer


Joined: 04 Jul 2012
Posts: 1551

PostPosted: Sat Oct 26, 2013 10:57 am    Post subject: Reply with quote

Well, then you could look for another service to use and configure that service to do it; these tasks you describe are tasks that should not belong in an init system. Why should something designed to start services know of the fine details of the location the user is at, touch files totally not related to itself or take the decision itself to mount a certain (encrypted) partition based on a set of rules. Of course you could easily do many of that in systemd, but why would you set up such a design that is the opposite of what most of us want from systemd (it being minimal); it's preferable to have the services do this, as their functionality is not supposed to merge into systemd.

If you do however want it; you can still write services that run scripts to start specific targets, or otherwise use /etc/local.d (for example, especially handy for mounting the encrypted partition).

But if you go down that road then you go back to the old days of hacking it in and giving up on the comfort that services do, can or should provide...
Back to top
View user's profile Send private message
mv
Advocate
Advocate


Joined: 20 Apr 2005
Posts: 4219

PostPosted: Sat Oct 26, 2013 8:35 pm    Post subject: Reply with quote

TomWij wrote:
these tasks you describe are tasks that should not belong in an init system.

The tasks do not, the possibility to start and configure tasks and kernel modules (based on user-defined conditions) does.
systemd fails to do this, since it has only a primitive static setup.
Quote:
you can still write services that run scripts to start specific targets

Yes:
Mask the networkmanager unit and write a script which is able to start networkmanager based on a dynamic configuration.
Mask the encryption unit(s) and write a script which is able to start encryption support based on a dynamic configuration.
Mask the loading of modules somehow (BTW: not even possible with systemd) and write a script which is able to load the modules based on a dynamic configuration.
etc. pp.
In other words: Switch off all the systemd and write a configurable init-system without systemd supporting you in any way.
As I said: The more reasonable way is to wipe systemd and use a configurable sane init-system from the very beginning (instead of writing your own init-system on top of the retarded static systemd BS).
Back to top
View user's profile Send private message
Display posts from previous:   
This topic is locked: you cannot edit posts or make replies.    Gentoo Forums Forum Index Gentoo Chat All times are GMT
Goto page Previous  1, 2, 3, 4, 5, 6, 7  Next
Page 4 of 7

 
Jump to:  
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