Hello, On Wed, 13 Aug 2014, Guido Berhoerster wrote: [..]

I'm really wondering what this is supposed to achieve now that everything already depends on systemd (particularly the middle stack on which the desktops now depends on logind which is tightly coupled to systemd as init,

There you have the core problem of systemd gobbling up stuff ... [..]

BTW, you could also do something constructive regarding systemd by reverting the switch from rsyslog to journald which causes absymal performance as notend several threads on this list.

That's what systemd-shim is supposed to solve along with eudev, cgmanager, etc. -dnh -- I'm telling you that the kernel is stable not because it's a kernel, but because I refuse to listen to arguments like this. -- Linus Torvalds -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Hello, I'll admit right away that I'm from the camp that really likes freedom of choice, modularity and a truly open and configurable system. I wasn't around for much of the history of this transition, so I have been following this thread with great interest... and must admit I've learned a great deal because of it. The remainder is in-line. On Wed, Aug 13, 2014 at 03:05:10PM -0400, Cristian Rodríguez wrote:

El 13/08/14 a las #4, Tomas Cech escribió:

...

Hi people,

to get some better overview about our systemd opinions, mainly, to quantify it I created a poll:

https://connect.opensuse.org/pg/polls/read/sleep_walker/45294/do-we-want-to-...

To whoever thinks this is a good idea (it is not) could you kindly answer the following questions ?

There are way more experienced people posting to this thread, but why not, I'll take a stab at a few of them... even at the risk of sounding like the dumbest person in the room.

- What you will use to replace logind ? (and who is going to maintain that replacement...)

This I don't know.

- What you will use when udev stops working without systemd ?

There are other udev forks... gentoo maintains one (eudev). Whether forking this is a good idea or not, I would let someone with greater udev knowledge answer.

- What you will use as a cgroup writer process ? (google: "single cgroup writer")

Tricky... cgmanager seems to be working for Debian at the moment. Though it seems as the cgroups API changes, this may no longer be the case. https://lists.debian.org/debian-ctte/2013/12/msg00149.html

- How you will convince various upstreams including but not limited to several freedesktop key components, plasma on wayland, gnome-session (?) to accept all the needed code that this effort necessarily entails ?

Also don't know.

- In the future.. what you will use to replace ..let's say.. avahi (aka. systemd-resolved) ?

If systemd-resolved is only doing /etc/resolv.conf setup, then our Wicked does this just fine through `netconfig`. The general tone I get from this discussion is that there are a lot of people, myself very much included, that like having options/choice. Seeing that choice diminished doesn't sit well with many of us. I think discussing possible ways to re-introduce at least some of this choice is very constructive and far from a bad idea.

I can use the rest of the evening writing a longer list of questions and detailing why this is a DOA idea for which you will find no support where it matters but I will just leave it there.

This ship sailed already and is in the middle of the ocean.. so if you want to be constructive (I doubt that.. but hey.. I might be wrong) you can start now by:

Pointing exactly what your actual, verifiable problems with systemd are and by that I do not mean vague, philosophical complains.

Actually I think many of the concerns brought up in this thread are far from vague, even the ones touching on philosophy. I see nothing vague about a philosophical stance for or against how an organization acts.

They are much more likely to be solved if:

* you post them to the systemd mailing list where it can get attention of the relevant people.

* Bonus points if you are absolutely sure it is a systemd problem and have taking the time to understand *why* and *how* things work the way they do.

Example on what will work:

- "I want to do X, but it systemd-Yprocess crashes"

- "I need X, but there is no option to do so."

- "The journal is slow and uses X,Y amount of CPU and RAM" (this is a known, multi-faced problem that some people have reported that needs to be fixed)

-- Cristian "I don't know the key to success, but the key to failure is trying to please everybody." -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Am 13.08.2014 um 21:05 schrieb Cristian Rodríguez:

El 13/08/14 a las #4, Tomas Cech escribió:

...

https://connect.opensuse.org/pg/polls/read/sleep_walker/45294/do-we-want-to-...

To whoever thinks this is a good idea (it is not)

I agree that this is not a good idea, but your questions are too funny (and too easy to answer) for me to skip the fun :-)

could you kindly answer the following questions ?

- What you will use to replace logind ? (and who is going to maintain that replacement...)

busybox

- What you will use when udev stops working without systemd ?

mdev

- What you will use as a cgroup writer process ? (google: "single cgroup writer")

Build a kernel with CONFIG_CGROUPS=n

- How you will convince various upstreams including but not limited to several freedesktop key components, plasma on wayland, gnome-session (?) to accept all the needed code that this effort necessarily entails ?

Well, probably it would be possible to just build *BSD userspace. There must be some software that runs without systemd. There are operating systems that do not use systemd :-) XFCE runs on BSD, so it is not necessary to have systemd for it to work.

- In the future.. what you will use to replace ..let's say.. avahi (aka. systemd-resolved) ?

I never had a use for avahi anyway, I am certainly not going to miss it.

I can use the rest of the evening writing a longer list of questions and detailing why this is a DOA idea for which you will find no support where it matters but I will just leave it there.

I could spend the rest of the evening writing "busybox" as an answer :-P

- "The journal is slow and uses X,Y amount of CPU and RAM" (this is a known, multi-faced problem that some people have reported that needs to be fixed)

This is my favorite bug, waiting for a solution since the first incantation of journald arrived. Now we have binary logs with abysmal performance and no program to read them. Too bad. :-) No matter how badly my mahchines crashed and how corrupted my /var/log/messages was, but there was always the possibility to skip over the bad spot in the logfile and read the messages before and / or after. The problem is: right now the forwarding to syslog-ng is broken again (even though the syslog-ng package did not change), so there is no reliable log at all on factory. -- Stefan Seyfried "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Richard Feynman -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

El 13/08/14 a las #4, Felix Miata escribió:

On 2014-08-13 15:05 (GMT-0400) Cristian Rodríguez composed:

...

- What you will use to replace logind ?

Hasn't it been broken from its inception? Is it or systemd itself not the underlying cause of https://bugzilla.novell.com/show_bug.cgi?id=833253 ?

PAM configuration problems have nothing to do with logind.. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Wednesday 13 of August 2014 15:05:10 Cristian Rodríguez wrote:

- What you will use when udev stops working without systemd ?

Funny. At the time udev was assimilated into systemd at the source tree level, our concerns wer being calmed down by claims that this is not going to happen. Now when you believe you have your final victory, you take it as a given thing that udev is going to be systemd-only. And that's exactly (one part of) the systemd attitude which makes me reject the project. Michal Kubeček -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

El 13/08/14 a las #4, David Haller escribió:

Hello,

On Thu, 14 Aug 2014, Hans-Peter Jansen wrote:

...

FYI:

https://lkml.org/lkml/2014/8/12/459

What a great summing up gem you found there! Thanks H-P!

"Implementing systemd by distros is not a wise move for them over the long term. It will, in fact, be their ultimate undoing. [...] Who are these people anyway? Who are these self-appointed keepers of the Linux flame? [...] systemd is scary - not just because it's tools suck, or because it's a massive fucking hairball - but because architecturally it has way too much concentrated power."

(I'm tempted to cite almost all of that mail, but that'd be, well .. ;)

From a follow-up https://lkml.org/lkml/2014/8/13/235:

"For any account: Depending on a particular init system in a desktop environment is a *bug*."

It is not a bug.. but who cares you are basing your comment from troll post to LKML and clearly do not have a single clue about how and why things work this way in the first place. -- Cristian "I don't know the key to success, but the key to failure is trying to please everybody." -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Thursday 2014-08-14 01:56, David Haller wrote:

PID1 can be implemented in ~15 lines of C.

With already 15 steps to start a daemon (man 7 daemon), definitely not. That is, unless you kill all the newlines in a C program and discount the preprocessor lines.

Linux is about choice. Has always been.

http://islinuxaboutchoice.com/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Thursday 14 of August 2014 03:48:59 Jan Engelhardt wrote:

On Thursday 2014-08-14 01:56, David Haller wrote:

...

PID1 can be implemented in ~15 lines of C.

With already 15 steps to start a daemon (man 7 daemon), definitely not. That is, unless you kill all the newlines in a C program and discount the preprocessor lines.

Without wanting to discuss if the original claim is right, you argument definitely isn't. Most of these 15 steps are not needed at all for root process and some would be even harmful (hint: think what happens if you follow step 15). Michal Kubeček -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Friday 15 of August 2014 07:17:41 Jan Engelhardt wrote:

On Friday 2014-08-15 07:04, Michal Kubecek wrote:

...

On Thursday 14 of August 2014 03:48:59 Jan Engelhardt wrote:

...

On Thursday 2014-08-14 01:56, David Haller wrote:

...

PID1 can be implemented in ~15 lines of C.

With already 15 steps to start a daemon (man 7 daemon), definitely not. That is, unless you kill all the newlines in a C program and discount the preprocessor lines.

Without wanting to discuss if the original claim is right, you argument definitely isn't. Most of these 15 steps are not needed at all for root process and some would be even harmful (hint: think what happens if you follow step 15).

What's critical about #15? Tell me, you seem to know, and since you do, openSUSE surely has patched it out and reported it to upstream, right?

When PID 1 finishes, the whole system goes down. Michal Kubeček -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Friday 15 of August 2014 07:44:08 Jan Engelhardt wrote:

On Friday 2014-08-15 07:19, Michal Kubecek wrote:

...

...

...

...

With already 15 steps to start a daemon (man 7 daemon), definitely not. That is, unless you kill all the newlines in a C program and discount the preprocessor lines.

Without wanting to discuss if the original claim is right, you argument definitely isn't. Most of these 15 steps are not needed at all for root process and some would be even harmful (hint: think what happens if you follow step 15).

What's critical about #15? Tell me, you seem to know, and since you do, openSUSE surely has patched it out and reported it to upstream, right?

When PID 1 finishes, the whole system goes down.

Note how "original process" does not refer to PID1, but "a traditional SysV daemon" that starts.

You already deleted the quoted part but the original claim you responded to was PID1 can be implemented in ~15 lines of C. To which you responded with the argument about 15 steps needed to start a deamon. But starting a daemon is not something root process has to do so those 15 steps are irrelevant for the question how much source you need to implement PID 1. Michal Kubeček -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Thursday 2014-08-14 09:56, Todd Rme wrote:

3. Require that package maintainers accept patches to add support for alternatives to packages that don't support it 4. Require that package maintainers accept submit requests to add support for alternatives without requiring patches 5. Provide the basic infrastructure to support alternatives and let package maintainers to decide what to do for their own packages. (of course 1 implies 2-5, 2 implies 3-5, 3 implies 4-5, and 4 implies 5)

What if a package maintainer accepts a sysv script, something changes upstream down the road, the sysv script or whatever needs to be substantially changed, and the person who submitted it has disappeared?

When upstream does not provide a file that is usable with a particular supported init system, we used to write our own (and continue to do so today) and ship them alongside the .spec file in the SRPM. Every addition of patches and {extra files not maintained by upstream} to a SRPM I hate to do and maintain, and hate it by line count. Therefore, for any alternative init system to halfway succeed, in openSUSE, you better make sure it can reuse existing .service files (at least a basic command subset). I am optimistic that there is such progress in some other init(s). -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Hey, On 14.08.2014 10:53, Tomas Cech wrote:

The poll was meant for simple quantifying the voices here to get the idea, what do we want.

I hope you realize that you make people think, that by casting their vote, something will happen.... Henne -- Henne Vogelsang http://www.opensuse.org Everybody has a plan, until they get hit. - Mike Tyson -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 08/14/2014 09:01 AM, Tomas Cech wrote:

On Thu, Aug 14, 2014 at 02:06:59PM +0200, Henne Vogelsang wrote:

...

Hey,

On 14.08.2014 10:53, Tomas Cech wrote:

...

The poll was meant for simple quantifying the voices here to get the idea, what do we want.

I hope you realize that you make people think, that by casting their vote, something will happen....

What do you expect to change by the clicking to poll besides expressing opinion?

Well depending on the "outcome" of the poll different expectations may arise. For example should the "we need an alternative" question garner a majority or even be a close second, people might get the expectation that "someone" will implement this alternative. Unfortunately it appears very common, even for people on this list, to think, "I voiced my opinion and a number of other people agreed, thus someone should go implement that." But that's not how it works. It goes back to the same old point, those that want an alternative need to implement the alternative and not expect other people to do the work for them. Objections to the "implement the alternative" approach are also always the same - - upstream makes changes to the code to depend on systemd - - there are too many packages to change - - the work is not accepted into Factory - - ..... - - excuses, excuses, excuses.... round and round it goes. Then everything dies off for a while and then we start again. But the bottom line is really quite simple. If an alternative is needed do the work to make it possible and then show everyone. This work will include not only distribution work, but also convincing people in upstream projects that the code should handle multiple init systems and participating in those communities to maintain that code. If enough people like the alternative they will start to contribute to the effort. Another option is of course to get involved in the systemd project and make commits there to shape it. To this people state that Kay and Lennard are evil incarnate and will not accept their patches. Well currently the number of contributors to systemd in github stands at 322. Thus it appears that patches do get accepted from a number of contributors. Of course the project has their philosophy and their acceptance criteria, that's no different than the kernel, glibc, gcc, you name it. Later, Robert - -- Robert Schweikert MAY THE SOURCE BE WITH YOU SUSE-IBM Software Integration Center LINUX Tech Lead Public Cloud Architect rjschwei@suse.com rschweik@ca.ibm.com 781-464-8147 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iQEcBAEBAgAGBQJT7LsOAAoJEE4FgL32d2UkxyQH/3GL4FEA3F4T1IQDjSpwo2sn VJCk+xbM2uPpnnZCvvAcfFw/7i1cbkJRJVvOC+ZLwl/G7s/296oS4oLg/MVRie5S ZsHw7YWpg3bHQ8EiUeY2WJtdGpGpjoGy11YcrNj8e7RTDQ3xqBhXdmwYc6eKgo6f +RsM6G4Hyhdh8g7tr0k3nwmcxmcpPlQgoU7+qCawLcVaViNRO4ikq+mfcSDnVyLB KiXmbD6Nmg5pf5bNqesZ92Cdk5fLo7yTgRkxaxSG4Ibz220b7FxixK/WBQtA/X1H cyhHMZbpAAIb6w0mNqZjAXfxUnojFy9ywBfArEexEKuOXXytjmRbkAfDIJdGzXA= =RWgt -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On 2014-08-14T16:06:48, Tomas Cech wrote:

...

Then everything dies off for a while and then we start again. I'm not aware that there would be any survey on this topic in the past. I may be wrong though.

That's been the result of every single systemd complaint. Noone just stepped up and maintained one of the alternative init systems. And frankly, systemd is not bad technology. It's certainly much better than sysvinit was, and the dependency model solves a number of issues for me, also the more event-driven approach. There are a huge number of excellent ideas in it. I really wish everyone who keeps complaining would be doing something more constructive.

If we want to contribute, we need to show that there is group of people interested in that. If we can show that, we can try to convince people to accept our modifications to packages.

You do show this by being persistent about the patches, keeping them maintained, and as self-contained as possible. Do by doing, not by polling.

Please, stop offering me participation in that project. That is not the point here.

So who is supposed to be doing the work you're asking for? Regards, Lars -- Architect Storage/HA SUSE LINUX Products GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) "Experience is the name everyone gives to their mistakes." -- Oscar Wilde -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Thu, Aug 14, 2014 at 04:35:36PM +0200, Lars Marowsky-Bree wrote:

...

If we want to contribute, we need to show that there is group of people interested in that. If we can show that, we can try to convince people to accept our modifications to packages.

You do show this by being persistent about the patches, keeping them maintained, and as self-contained as possible. Do by doing, not by polling.

You're describing fork of distribution packages.

...

Please, stop offering me participation in that project. That is not the point here.

So who is supposed to be doing the work you're asking for?

Maybe I poorly formulated my response. Don't offer me working on improving systemd. S_W

Y0, On 14.08.2014 16:06, Tomas Cech wrote:

On Thu, Aug 14, 2014 at 09:35:10AM -0400, Robert Schweikert wrote:

...

On 08/14/2014 09:01 AM, Tomas Cech wrote:

...

On Thu, Aug 14, 2014 at 02:06:59PM +0200, Henne Vogelsang wrote:

...

On 14.08.2014 10:53, Tomas Cech wrote:

...

The poll was meant for simple quantifying the voices here to get the idea, what do we want.

I hope you realize that you make people think, that by casting their vote, something will happen....

What do you expect to change by the clicking to poll besides expressing opinion?

For example should the "we need an alternative" question garner a majority or even be a close second, people might get the expectation that "someone" will implement this alternative.

I'm sorry but I don't believe this is the case.

Fair enough, let's see.

Result of such vote, if it would be successful and taken seriously would be that "someone" _can_ implement alternative into openSUSE.

What has the poll to do with this? If you, I or anyone else wants to do this today, we can. Even if EVERYBODY votes against this in your poll... Henne -- Henne Vogelsang http://www.opensuse.org Everybody has a plan, until they get hit. - Mike Tyson -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On 2014-08-14 19:34 (GMT+0200) Tomas Cech composed:

This poll should show, how much people are interested in this idea. So patch submission won't be rejected by package maintainer for other than technical reasons. Vox populi, vox dei...

How many votes were lost before the reset, and at what ratio? -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Moin, On Thu, 14 Aug 2014, 15:35:10 +0200, Robert Schweikert wrote:

[...] But that's not how it works. It goes back to the same old point, those that want an alternative need to implement the alternative

yeah, and that's exactly how projects like "kernel, glibc, gcc, you name it" got where we are now

and not expect other people to do the work for them. Objections to the "implement the alternative" approach are also always the same

- - upstream makes changes to the code to depend on systemd

To be honest, looking at how deep systemd got injected into all sort of public projects, we should be prepared to accept maintaining potentially several of such pieces on our own, at least for a period of time.

- - there are too many packages to change

Yes, we need to get a plan what to touch and when...

- - the work is not accepted into Factory

then let's create our Factory_without_sytemd rolling release ;)

- - ..... - - excuses, excuses, excuses....

not accepted ;) We've got a common vision and goal, let's go for it. Cheers. l8er manfred -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org