Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
When (and if) Gentoo will switch to systemd?
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3 ... 6, 7, 8 ... 16, 17, 18  Next  
Reply to topic    Gentoo Forums Forum Index Other Things Gentoo
View previous topic :: View next topic  
Author Message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Tue Feb 11, 2014 12:45 am    Post subject: Reply with quote

Anon-E-moose wrote:
If systemd were to be setup this way.

Core libs - that all parts use

Modules that I can compile separately using core
init
logging
cron
login
other stuff

each not depending on compiling everything, then I would say that it was modular.
Putting each file, or include in its own directory does not make systemd modular.
Each file or subsystem may be modular, from a source standpoint, but that does not make systemd modular.


+1

Generally speaking if code can't be separated out of its environment and tested in a unit test, and subsequently be joined together and easily tested in an integration test, then it's probably not very well written. Everything I see about systemd says it needs all sorts of extras (like the whole ball of wax, and then dictating what sort of utilities I use) which pretty much says it can't be separated out very well.

Redhat has lots of resources. Rather than trying to force a bad design down everyone's throats, why not fix it and make a few independent services that could be mixed and matched, the way standard UN*X works? There is no good reason for all these interdependencies. Why not just make an additional API (or a few) that each thing implements, have each tool check for the implementation of that api and go from there? That way people can pick and choose.
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Tue Feb 11, 2014 1:09 am    Post subject: Reply with quote

1clue wrote:
Redhat has lots of resources. Rather than trying to force a bad design down everyone's throats, why not fix it and make a few independent services that could be mixed and matched, the way standard UN*X works? There is no good reason for all these interdependencies. Why not just make an additional API (or a few) that each thing implements, have each tool check for the implementation of that api and go from there? That way people can pick and choose.


I agree with what you say about RH and part of the reason that I just don't cotton to the idea of going with systemd.
If they had decided to implement something like sysvinit/openrc and stopped at that, I could see the benefit of it.

But I'm happy with logging, cron, login service, etc the way it is and don't want an all in one solution.
I already have windows boxes around the house if I need that type software/mentality.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
krinn
Advocate
Advocate


Joined: 02 May 2003
Posts: 3921

PostPosted: Tue Feb 11, 2014 10:01 am    Post subject: Reply with quote

1clue wrote:
@ovitters,

It seems you joined the forum simply to post on this thread, since you joined today and as yet all 3 of your posts are on this thread. It also seems that you are this guy:
http://blogs.gnome.org/ovitters/author/ovitters/

If you are that guy, then I'm going to make a wild leap and say you're here because of the politics of the situation.


And if you are that guy, can you edit your blog to stop your false propagenda !
http://blogs.gnome.org/ovitters/author/ovitters/ wrote:
Loads of distributions have switched to systemd. Especially the distributions with a lot of people behind them (paid or not). Now even Gentoo is ok with a systemd dependency. On Gentoo the packagers added a systemd dependency since a few months ago.

Now even Gentoo is ok with systemd... With a link to http://thread.gmane.org/gmane.linux.gentoo.devel/88551

Look at Pacho Ramos (the guy that initiate that news) blog quote : http://my.opera.com/pacho/blog/2013/07/24/gnome-3-8-requiring-systemd-on-gentoo
Quote:
After spending a lot of time trying to let logind run without systemd ..., all the efforts failed because systemd-205 no longer let us run logind alone as cgroups handling was moved to systemd itself


Gentoo Gnome Team is not ok with it ; it's a surrender nothing more.
And even if Gentoo Gnome Team was happy with it, in no way it means Gentoo itself is ok with it. How can you say a distro that its first power is base on use flags could be "ok" with any program that have hard dependency ; weakening Gentoo power over other distros.

I'm not ok with your propaganda on debian users, but i won't goes pollute debian forum with my position as i don't use debian myself (even i personally have good feeling on debian) ; but don't use Gentoo to backup your propaganda over debian users with false claims (leave that to Pottering).
Back to top
View user's profile Send private message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Tue Feb 11, 2014 3:34 pm    Post subject: Reply with quote

And for that matter, why is it that Gnome seems to think that it's the only such software in Linux?

I don't use Gnome, especially not the latest stuff.

I don't mind systemd being in the repository, but if it suddenly becomes required in order to run Gentoo, then I'm going to find some other distro. Some of my source-based installs are aiming at extremely tiny and extremely simple installs. Why would I want systemd in there?
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Tue Feb 11, 2014 4:58 pm    Post subject: Reply with quote

I think most of us who don't use systemd don't mind it being in the repository.

If someone wants to use it, I won't talk them out of it. It's their choice.
I make my own choices though, and for my purposes,
I see absolutely no need for systemd on any of my systems.

I have gtk2 for a handful of programs, and gtk3 (3.4.4) for exactly 2 programs usbview and zathura/girara.
I also have only 2 gnome libraries, libglade and librsvg, they're old but they work for my uses
and if someone were to put out a non-gnome version of them that would be nice. Choice is good, IMO.

I refuse to keep up with the continually breaking madness of gtk3 so I've halted it where it is.
If a program needs a later version, I will simply find another program whether qt based or something else.

I use lxde which uses gtk2 at the moment, but I'll probably switch when the razor-qt merge matures a touch.

I'm not an anti-gnome-gtk-zealot, but I see limited to no use for them with the direction they are heading.
I wish them well, along with all those who choose to travel that path. It's just not for me.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
rudregues
Apprentice
Apprentice


Joined: 29 Jan 2013
Posts: 188

PostPosted: Tue Feb 11, 2014 8:39 pm    Post subject: Reply with quote

Anon-E-moose wrote:


I use lxde which uses gtk2 at the moment, but I'll probably switch when the razor-qt merge matures a touch.


In future lxde will use qt. No gtk3 will enter
_________________
Emerging en gentoo
Back to top
View user's profile Send private message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Wed Feb 12, 2014 12:05 am    Post subject: Re: When (and if) Gentoo will switch to systemd? Reply with quote

mv wrote:
Anon-E-moose wrote:
iptables

It seems that I am missing some information: Why do you think that iptables (or its successor nftables?) is tied to the systemd project?


I think he's speculating that it will come to that.

After reading some of ovitters' blog I think he (Anon-E-moose) is right. Systemd is the Borg. Resistance is futile: You will be assimilated.

This project is being handled in an extremely nonstandard way.

Usually somebody (some group) comes up with a new package, works the bugs out of it and slowly gains traction as people switch over. Usually for any specific functionality there are a couple really good options and a handful of decent ones that might appeal to edge cases. Everyone here knows this.

Redhat and Gnome are trying to ram this thing down everyone's throats, and doing their best to make it the only thing that can be used on any system. They're doing it by causing dependencies on a specific init system, which is just not rational to me. I feel like I'm on a used car lot and somebody's just a little too eager to sell me a specific car that I don't really want.

This thing is infectious, monolithic and heavy. I have several use cases where having to use systemd would hurt my end result.

Why can't systemd split off plugins? Make JUST an init system that's aware of extra functionality but doesn't have to use it, and then add the rest of the "magical" functionality as wanted? There's zero good reasons for the /usr merge. There's zero reason for a dependency on a specific init system.

The ENTIRE POINT of running Linux is to use what you feel necessary and not be shoehorned into something that doesn't fit your use case. The ENTIRE POINT of Gentoo is to be able to choose the things you feel best suit your need, even to the point of compile options.

If systemd is supposed to be so great, then why are its proponents pushing so hard to force it on us? Sit back and let the code do the talking.
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Wed Feb 12, 2014 1:04 am    Post subject: Reply with quote

Speculation on my part, but RH/LP want a windows like system, and since they can't co-opt the kernel,
they are building something on top of it, similar to the early versions of windows that sat on top of dos.
But what everyone will see is a "shiny front end". Or at least most desktop users.

In the short time I've known of systemd, I've seen it grow by leaps and bounds with it's kitchen sink philosophy.
With dbus interface kdbus "just having to be in the kernel".
So I don't think they will stop until they have all the pieces to "challenge" windows. :roll:

Edit to add: The kickback against ubuntu when they tried the tablet interface on a desktop,
as well as the general lack of enthusiasm for windows 8 will make it a real uphill struggle
to get gnome 3 accepted by corporations. And if enough desktops standardize on systemd,
I think they will find that the script kiddies have an awakened interest in finding all the holes
in their whiz-bang software.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Wed Feb 12, 2014 3:33 am    Post subject: Reply with quote

I respectfully disagree.

They CAN co-opt the kernel. Google did it with Android, the source is there, all they gotta do is do it.

The ONLY problem with that is that there's no way the official kernel guys would pull the changes back in to mainstream, so systemd would be limited to redhat-only systems. Which is where it should stay, the more I look at it.

Seriously I don't think Redhat or anyone else is going to seriously compete with Windows on the open market. I frankly don't care. My use case for Linux is to be what Windows isn't, so why try to duplicated it? And for that matter, Redhat is aimed mostly at the server market, where it makes sense.
Back to top
View user's profile Send private message
miket
Apprentice
Apprentice


Joined: 28 Apr 2007
Posts: 173
Location: Gainesville, FL, USA

PostPosted: Wed Feb 12, 2014 4:47 am    Post subject: Reply with quote

1clue wrote:
They CAN co-opt the kernel. Google did it with Android, the source is there, all they gotta do is do it.

Why, Red Hat's already done that kind of thing. They put, so help me, a HTTP server in the kernel.
From the Wikipedia article:
Quote:
TUX has never been an integrated part of the official Linux kernel, although it has been shipped in some distributions, notably Red Hat, SuSE and Fedora.

From what I can tell, that patch was only in 2.4 kernels. I sure hope it didn't go any further!
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Wed Feb 12, 2014 10:38 am    Post subject: Reply with quote

1clue wrote:
I respectfully disagree.

They CAN co-opt the kernel. Google did it with Android, the source is there, all they gotta do is do it.

The ONLY problem with that is that there's no way the official kernel guys would pull the changes back in to mainstream, so systemd would be limited to redhat-only systems. Which is where it should stay, the more I look at it.


Well, that's what I meant when I said co-opt. Linus won't let them take over the mainstream kernel.
They can't make all of linux's kernel be RH's version.
Oh they may make their own mods to the mainline kernel, AFAIK they do that today via patches.
But as you said it's RH's version, not linux per se. Unless I grab their patches and apply it to the
kernel that I get from kernel.org, I'm not running RH's patched version of linux.

Quote:
Seriously I don't think Redhat or anyone else is going to seriously compete with Windows on the open market. I frankly don't care. My use case for Linux is to be what Windows isn't, so why try to duplicated it? And for that matter, Redhat is aimed mostly at the server market, where it makes sense.


I don't think they'll compete either, but RH has been lusting after that market for a long time.
RH is trying to be all things to all people. Sound familiar? :lol:

I agree that RH's money is being made in the server market, which doesn't make sense that they
will try and shove systemd down the throats of the server buyers.
I've worked at too many large companies, and they don't like massive changes like systemd
inspired madness when they're not sure it will be of benefit to them. Hell with even minor
upgrades to AIX or other UNIXs they take a long time to certify that their machines will continue
to work the way they want them, 99.99% uptime with the same day to day reliability.
And systemd doesn't have that kind of testing as it's more geared to the desktop market and
relatively new. So I expect that RH will have a two pronged approach, systemd for desktop
and the servers to stay the way they are. If not, I think when companies do get to the point of
upgrading to something resembling systemd they may look at other things. It's not that hard
to switch to bsd machines. Companies don't run servers on the bleeding edge of software, unlike desktop users.

Anyway...just some rambling as I don't really care what RH does or doesn't do.
To me they've been irrelevant for quite a few years.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
miket
Apprentice
Apprentice


Joined: 28 Apr 2007
Posts: 173
Location: Gainesville, FL, USA

PostPosted: Wed Feb 12, 2014 3:23 pm    Post subject: Reply with quote

Anon-E-moose wrote:
I agree that RH's money is being made in the server market, which doesn't make sense that they
will try and shove systemd down the throats of the server buyers.

What a possibility for a perverse outcome! Red Hat, by promoting systemd on the desktop and, by this, managing to push Debian to systemd, may gaming to push system administrators away from Debian. Where? How about to a non-systemd server edition of RHEL?

So far as I know, Debian will continue to offer OpenRC, Upstart, and maybe their own stock SysVInit as options--if for no other reason to keep server admins on board. So long as a distro that big has to go through the effort of supporting multiple init systems, we are not as isolated as we might be.
Back to top
View user's profile Send private message
Shamus397
Tux's lil' helper
Tux's lil' helper


Joined: 03 Apr 2005
Posts: 126
Location: Ur-th

PostPosted: Wed Feb 12, 2014 3:53 pm    Post subject: Reply with quote

@miket: Did you hear that from their mailing list? Have they made their decision public yet? :)

If that's the case, that's great news! :D
Back to top
View user's profile Send private message
miket
Apprentice
Apprentice


Joined: 28 Apr 2007
Posts: 173
Location: Gainesville, FL, USA

PostPosted: Wed Feb 12, 2014 5:17 pm    Post subject: Reply with quote

Shamus397 wrote:
@miket: Did you hear that from their mailing list? Have they made their decision public yet? :)

If that's the case, that's great news! :D

I'm afraid it's only speculation. Despite the announcement on LWN about the Debian Technical Committee's vote, the issue still seems somewhat murky. My bet, judging from Debian's conservative nature, their large userbase of server administrators, and their desire to support HURD and BSD, is that they will continue to support other init systems. The Debian admins I know are alarmed about having systemd thrust upon them. It would mean major changes with little certainty.
Back to top
View user's profile Send private message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Wed Feb 12, 2014 5:28 pm    Post subject: Reply with quote

Anon-E-moose wrote:
Anyway...just some rambling as I don't really care what RH does or doesn't do.
To me they've been irrelevant for quite a few years.


I'm with you on this, unless they start playing in my sandbox.

Ubuntu is talking about upstart and systemd, but it's not as urgent a discussion there because almost nobody knows what you're talking about.
Back to top
View user's profile Send private message
depontius
Advocate
Advocate


Joined: 05 May 2004
Posts: 2384

PostPosted: Thu Feb 13, 2014 1:45 pm    Post subject: Reply with quote

In the spirit of keeping this thread technical, some interesting reading. I haven't actually gone through it all yet, but plan to. It's a non-init-related developer looking at making his stuff work with systemd, along with legacy comments on SysV.

Systemd programming part 1: modularity and configuration https://lwn.net/Articles/584175/
_________________
.sigs waste space and bandwidth
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Thu Feb 13, 2014 3:29 pm    Post subject: Reply with quote

I do find it mildly amusing to watch the ever changing technically superior reasons for using systemd.

At first it was faster bootup times, but when that got pretty much demolished, as not as true as was
stated or in the case of large companies, irrelevant, then it became something else.
And that morphing hasn't stopped yet. At glancing at the article the one superiority
brought out that it is "easy" to override defaults in booting up.

Sometimes I'm just easily amused.

Thanks for the article, though.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
depontius
Advocate
Advocate


Joined: 05 May 2004
Posts: 2384

PostPosted: Thu Feb 13, 2014 4:39 pm    Post subject: Reply with quote

I went back and read TFA over lunch... This little line struck me, and I came up with a neat way to inject hate & discontent into the discussion, without personal attacks or insults.

Code:
There are at least two possible responses that are worth considering to address this need. The first is to modify all the daemons to accept any command line configuration also from environment variables. For example mountd could be changed to examine the MOUNTD_PORT environment variable if no -p option were given, and use that value instead.


The solution is obvious - have the daemons get such parameters from dbus! Somewhere in this wide-ranging discussion I saw something about command-line initiated programs that don't read command-line parameters. They just fork into the background and start listening to dbus. The only way to "exert your will" upon them is through dbus - or "kill".

In a similar way I've found myself unable to use Adobe Digital Editions under Linux. I have it installed on one machine, but every aspect of it seems to be desktop-centric, and it doesn't appear to make it possible to run on the command line, or without one of the heavyweight desktops. So I'm not checking out library ebooks for my Kobo, or buying any ebooks for anyone but Tor or Baen.
_________________
.sigs waste space and bandwidth
Back to top
View user's profile Send private message
Shamus397
Tux's lil' helper
Tux's lil' helper


Joined: 03 Apr 2005
Posts: 126
Location: Ur-th

PostPosted: Thu Feb 13, 2014 6:23 pm    Post subject: Reply with quote

I find it very interesting that Martin Gräßlin, the maintainer of kwin, is saying that he's going to tie kwin to systemd. How convenient! Now, instead of just Gnome pushing systemd on the world, we'll now have KDE pushing too! :P The arguments he uses to justify it seem to be pretty weak too, those being that kwin needs cgroups and socket activation.

What's funny is the cgroups argument was debunked by Linus Torvalds of all people, back when Mr. Poettering was saying that the kernel devs didn't need to patch anything in the kernel, all everyone had to do was put some stuff into their .bashrc (link here). The socket activation issue? Sounds like a whole lot of hot air, but I don't know enough about it to have a real opinion. ;) Seems to me that you shouldn't have to rewrite an init system + lots of other bits and pieces in order to have it, but that's just me. :P
Back to top
View user's profile Send private message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Thu Feb 13, 2014 8:01 pm    Post subject: Reply with quote

Shamus397 wrote:
Seems to me that you shouldn't have to rewrite an init system + lots of other bits and pieces in order to have it, but that's just me. :P


Exactly.

I'm waiting for JUST a rewrite of an init system. I don't deny that it could take some overhauling, but I don't like any of the new solutions. Most of all I think the /usr merge is horse doo-doo, and the forced integration thing.

If one of these systems were actually better in some measurable way without some insane disadvantage I wouldn't be so incredibly against it.
Back to top
View user's profile Send private message
1clue
Veteran
Veteran


Joined: 05 Feb 2006
Posts: 1199

PostPosted: Thu Feb 13, 2014 8:02 pm    Post subject: Reply with quote

And with regards to KDE and Gnome both adopting systemd, all I can say is maybe it's time to revisit FVWM? I loved that thing back in the day.
Back to top
View user's profile Send private message
miket
Apprentice
Apprentice


Joined: 28 Apr 2007
Posts: 173
Location: Gainesville, FL, USA

PostPosted: Thu Feb 13, 2014 8:08 pm    Post subject: Reply with quote

Shamus397 wrote:
I find it very interesting that Martin Gräßlin, the maintainer of kwin, is saying that he's going to tie kwin to systemd. How convenient! Now, instead of just Gnome pushing systemd on the world, we'll now have KDE pushing too! :P The arguments he uses to justify it seem to be pretty weak too, those being that kwin needs cgroups and socket activation.

Is the whole world going absolutely crazy? That fact that Gnome is married to systemd was not any great concern of mine: I parted ways with Gnome more than ten years ago because I couldn't stand Gnome 2.

KDE has been a refuge--and in the init-system battle even more so because of their very conscious aims for being cross-platform. Actually, I've been working for ways--on and off--to get a KDE system complete with notifications, disk mounting, and power control but not using Consolekit or polkit. (I had partial success.)

Now we have someone with misplaced enthusiasm for systemd. I've been looking to the development of Wayland because I see lots of promise in it--especially since they state the desire *not* to take over the world. There is nothing about it that would seem to require cgroups or socket activation.

The semantic-desktop thing notwithstanding, I have found the KDE people to be pretty sane. Concerning Martin Gräßlin hiself, I've found a mixed bag. While he has sometimes come up with strange ideas, he was also the one to be particularly attentive to some problems I had experienced.


Shamus397 wrote:
Seems to me that you shouldn't have to rewrite an init system + lots of other bits and pieces in order to have it, but that's just me. :P

Systemd is supposed to be a great boon for daemon maintainers, but for cross-platform servers like Apache, MySQL, Postgres, OpenSSH, and BIND, having to have special handovers of open sockets from that single init system must be a special pain to implement.
Back to top
View user's profile Send private message
steveL
Advocate
Advocate


Joined: 13 Sep 2006
Posts: 2161
Location: The Peanut Gallery

PostPosted: Thu Feb 13, 2014 8:10 pm    Post subject: Reply with quote

Shamus397 wrote:
I find it very interesting that Martin Gräßlin, the maintainer of kwin, is saying that he's going to tie kwin to systemd. How convenient! Now, instead of just Gnome pushing systemd on the world, we'll now have KDE pushing too! :P The arguments he uses to justify it seem to be pretty weak too, those being that kwin needs cgroups and socket activation.

Yeah, unfortunately I don't think the new cgroups model (only allowing one cgroup manager) is as capable as the old, which simply exposed the knobs and left it up to the admin or userspace with appropriate privilege, to tweak them (and do nothing by default). OFC if all you want is to tie in the desktop to something else doing the work, that's fine; I just think it's a shame.
Quote:
What's funny is the cgroups argument was debunked by Linus Torvalds of all people, back when Mr. Poettering was saying that the kernel devs didn't need to patch anything in the kernel, all everyone had to do was put some stuff into their .bashrc (link here).

Interesting thread, thanks. Torvalds clearly intended both mechanisms (ie automate no-brainers, but leave the rest to the admin/userspace) and not in a crippled sense: the existing interface was designed along Unix lines (expose the mechanism, and leave the policy to the user.) I found his debunking of system daemons generally most amusing; the arrogance of Poettering's reply ("If you truly believe X, then I can't help you.") is staggering. It does sound like a delete_on_release is needed, alongside notify_on_release, though. (lost the article link, sorry, must be deep in thread.)
Quote:
The socket activation issue? Sounds like a whole lot of hot air, but I don't know enough about it to have a real opinion. ;) Seems to me that you shouldn't have to rewrite an init system + lots of other bits and pieces in order to have it, but that's just me. :P

Yeah; though there is a case for merging xinetd socket-activation with monitoring (like monit or runit?) so for instance there's nothing wrong with adding such a process to eg openrc: it just doesn't need to engulf the rest of the system. And if you're doing that, you can handle cgroups and dbus events if you want. You don't need to run as pid 1 in order to get children reparented: Sievers submitted a patch that Cox accepted for that a couple of years ago.

That's really all that was needed, apart from a broadcast-capable message queue: the natural implementation of that would use shared memory (QNX is the precedent, from the early 90s) but again it can be developed with the tools at hand for POSIX, if you just want to get it done, and make the case for a kernel impl. Or write a lib that wraps a localhost socket, as recommended by the kernel networking people when the discussion about it first came up. The point is you can do those fairly portable mechanisms and also add a better capability in-kernel for platforms like Linux, which in turn can be used as a building block for all sorts of applications, not just one.

I don't have any problem with the idea of systemd: the implementation sucks though, because of basic design flaws. The concern is that there's leakage into the kernel, with bastardisation of cgroups for all of us, and single-app mechanisms that are the opposite of flexible, when the alternative is so much simpler to do and so much more useful across the board.

What I find extraordinary is the way the fact that it's FLOSS means somehow we can't critically analyse and discuss a programming project, unless we actually work on it. Even if the process for the latter weren't blocked by self-appointed commercial gatekeepers, as a programmer I find it completely bizarre. It's like saying a doctor can't tsk at a badly-stitched wound, unless he is working in the same hospital as the person who carried it out, or a plumber can't criticise a botch-repair unless he was working on the same site.

Ludicrous; and if you tolerate that, you'll get the software you deserve.
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Thu Feb 13, 2014 8:13 pm    Post subject: Reply with quote

Shamus397 wrote:
The arguments he uses to justify it seem to be pretty weak too, those being that kwin needs cgroups and socket activation.


Interesting in that with the rewrite of cgroups in the kernel, call it cgroups ver 2 there is less need
for something like systemd to manage them. They will have a well defined interface. So I expect
to see a few packages that will do that very thing.

Quite frankly the whole cgroups argument is just as weak as the bootup time argument.
In the scheme of things there are very few that use cgroups.
And certainly none, that properly using nice wouldn't do effectively the same thing, ie kernel priority.

I don't really care as I never have like KDE, no matter which version. I left gnome early on in version 2.

I'm happy with other choices, and don't miss the bloat of either of the heavyweight packages.
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
Anon-E-moose
Advocate
Advocate


Joined: 23 May 2008
Posts: 2096
Location: Dallas area

PostPosted: Thu Feb 13, 2014 8:23 pm    Post subject: Reply with quote

steveL wrote:
I found his debunking of system daemons generally most amusing; the arrogance of Poettering's reply ("If you truly believe X, then I can't help you.") is staggering.


I am reading that link right now and thanks for it.

One thing came to mind as I was following through it, LP doesn't want the kernel to do any thing he can do in his software.
The impression I get is he's scared to death that if it gets in the kernel, that's one more reason not to use his software.

:lol:

Edit to add:
LMAO over how incompetent a debater LP is when others are a lot more technically informed than he is.

Quote:
How about you stop bringing *your* narrow definition of "desktop" to
the discussion? I am a kernel hacker but that doesn't mean I'm not
also a desktop user. I shouldn't need to play with cgroups from
userspace to keep music playing while I compile kernels. Things should
just work.


Reply to LP's trying to redefine desktop to his narrow viewpoint. :lol: Best read I've had in a few days.

What's so funny about the whole thread is reading the replies from the systemd supporters trying
to make sure that they are the only thing that can do these things. They don't want Linus and the
kernel involved at all. That Linus is getting involved tells me that he's not impressed with systemd's
trying to control things. An interesting power play.


And here you go in LP's own words
Quote:
The plan with systemd is to make it manage both the system and the sessions.

And if anyone thinks that he means to stop with the piece they were discussing, I got a bridge to sell you. :wink:


What I would love to see is the kernel to do most of what systemd re: cgroups is trying to accomplish
from userspace and the kernel devs give a flying finger to the LP groupies. :lol:
_________________
Asus m5a99fx, FX 8320 - amd64-multilib, 3.9.1-zen, eudev
glibc-2.17, gcc-4.7.3-r1, xorg-server-1.15, lxde, nouveau, oss4
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Other Things Gentoo All times are GMT
Goto page Previous  1, 2, 3 ... 6, 7, 8 ... 16, 17, 18  Next
Page 7 of 18

 
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