A non-programmer's hypothesis "confirmed"

Story: systemd row ends with Debian getting forkedTotal Replies: 38
Author Content
Ridcully

Dec 01, 2014
3:08 AM EDT
Several weeks ago, on an earlier thread concerning systemd, I put forward a hypothesis that systemd was a naked grab for power by Red Hat over the future of Linux using Poettering. My hypothesis was, by implication of the comments, "pooh-poohed - not possible - nothing like that".

I take absolutely no satisfaction in noting that more and more comments on this site are more or less falling into line with that original hypothesis and pretty much confirming it. Sure, as an ultimate user, I suspect systemd's presence will make little difference to me...but I definitely do care about what I personally perceive is the destructive effect systemd is having on the Linux world. Resignations, forking......what more do we need to indicate that Unix traditions and engineering are being thrown out with the bathwater ? I don't know the answers either, but I am very, very worried. And so far, unfortunately, I cannot see any major reason to alter my "intestinal feeling about the matter".
me1010

Dec 01, 2014
8:07 AM EDT
@Ridcully:

Your observation is known as confirmation bias.

https://en.wikipedia.org/wiki/Confirmation_bias

Quoting:Confirmation biases can be used to explain why some beliefs persist when the initial evidence for them is removed.[45] This belief perseverance effect has been shown by a series of experiments using what is called the "debriefing paradigm": participants read fake evidence for a hypothesis, their attitude change is measured, then the fakery is exposed in detail. Their attitudes are then measured once more to see if their belief returns to its previous level.

A common finding is that at least some of the initial belief remains even after a full debrief.[46] In one experiment, participants had to distinguish between real and fake suicide notes. The feedback was random: some were told they had done well while others were told they had performed badly. Even after being fully debriefed, participants were still influenced by the feedback. They still thought they were better or worse than average at that kind of task, depending on what they had initially been told.


It is not backed up with actual data, just with unsubstantiated opinions expressed by other believers in the cause. Succumbing to confirmation bias is a clear indication of acceptance of the extraordinarily poorly constructed systemd conspiracy theory.
jdixon

Dec 01, 2014
8:46 AM EDT
> Your observation is known as confirmation bias.

No, his observation may be influenced by or the result of confirmation bias (as have been all observations of every person everywhere throughout history), but that's not the same thing.

> It is not backed up with actual data, just with unsubstantiated opinions...

Because the opinions of the pro-systemd folks are oh so substantiated.

> Succumbing to confirmation bias is a clear indication...

Of being human, since all humans do it. Anyone who claims they don't is lying.

> ... of acceptance of the extraordinarily poorly constructed systemd conspiracy theory.

Yep, all systemd opponents everywhere think systemd is some great conspiracy, rather than the acts of a single company taking steps to meet what they view as their customer needs. I guess we must all be part of some vast anti-systemd conspiracy.

Hint, the problem is that the needs of Red Hat's customer base aren't the needs of all Linux users everywhere.
me1010

Dec 01, 2014
9:03 AM EDT
@jdixon:

Occam's Razor applies. Whether one is pro or anti systemd, the likelihood of a RH, Microsoft, [fill-in favorite evil guy] conspiracy to take over all of GNU/Linux is extremely small, especially in light of the factual evidence that:

1. systemd is FOSS, so direct competitors may use the program or not, may change the program or not, and may offer the program themselves repackaged.

2. systemd's code base has not actually absorbed more than a couple of smaller projects.

3. Nearly all 'dependencies' are optional and are merely incorporated in certain desktop applications to support an 'if it's there' optional behavior.

4. POSIX and *nix philosophy is sort of a joke... or Platonic if you want... as there is no 'perfect' adherent *nix distro either in the commercial world or in the FOSS world.

Considering very valid and real points 1 through 3 and my historically based but nevertheless opinion in 4... systemd conspiracy theories should go the way of the Dodo bird.

EDIT:

Oh! I wrote before reading everything in its entirety... sorry...

Quoting:the acts of a single company taking steps to meet what they view as their customer needs


Yes! That's exactly what I think as to why systemd exists. However, its wider adoption by other distros is not a RH conspiracy rather an acceptance of something that is 'better' than what was before -- maybe not precisely the 'best' fit for all purposes... but who ever said the 'best' product always wins the marketplace?
hughesjr

Dec 01, 2014
11:25 AM EDT
@jdixon

Neither is unity on Ubuntu, so people forked it.

Linux is open source.. So fork it and do what you want. Or use the fork these guys create. Then we can see who wins in the end.

Things change. Use what is good for you.
hughesjr

Dec 01, 2014
11:34 AM EDT
I mean, Red Hat backports security updates in RHEL and you don't see Debian doing that.

Other distros take the pieces they want from Red Hat or Ubuntu or Debian. No one is forcing anyone else to use systemd.. They are choosing to. It's open source for people to use if they choose. Don't like it, use something else.

Cinnamon and Mate exist because some distros don't want Gnome3. Libre Office exists because some don't want Open Office. mariadb and mysql, etc.
vainrveenr

Dec 01, 2014
1:05 PM EDT
Quoting:No one is forcing anyone else to use systemd.. They are choosing to. It's open source for people to use if they choose. Don't like it, use something else.


Quote from The Register piece:
Quoting:Negotiations have, of late, considered making systemd optional, but those talks appear not to have gone well if this post are anything to go by: it announces a fork called Devuan.


Quote from last week's post, linked from The Register snippet above:
Quoting:Subject: [Dng][debianfork] Don't panic and keep forking Debian™! Dear Init-Freedom lovers,

The Veteran Unix Admin collective salutes you.

As many of you might know already, the Init GR Debian vote promoted by Ian Jackson wasn't useful to protect Debian's legacy and its users from the systemd avalanche.

This situation prospects a lock in systemd dependencies which is de-facto threatening freedom of development and has serious consequences for Debian, its upstream and its downstream.

The CTTE managed to swap a dependency and gain us time over a subtle install of systemd over sysvinit, but even this process was exhausting and full of drama. Ultimately, a week ago, Ian Jackson resigned https://lists.debian.org/debian-ctte/2014/11/msg00091.html.
(Source is [Dng][debianfork] Don't panic and keep forking Debian™!)



Long story made short is that earnest attempts to make systemd "optional" for Debian Jessie have failed, so claims similar to the above that "No one is forcing anyone else to use systemd" could be considered spurious --- for Debian Jessie currently.

Hence a perceived need for the promising fork of Debian Jessie, Devuan, as the above source reveals, and which has become the very subject of this particular thread.





me1010

Dec 01, 2014
1:21 PM EDT
@vainrveenr:

...misleading...

Negotiations within Debian ... appear not to have gone well...

Neither RH nor systemd maintainers as individuals forced Debian to do anything.

Debian could have chosen to stay with sysvinit, move to openrc, adopt upstart, or any number of things. The politics of Debian are solely the realm of Debian and have little to do with the politics elsewhere.

Here are the choices considered:

https://lists.debian.org/debian-ctte/2014/02/msg00402.html

Again...

Quoting:Please check sources before posting incorrect information.
linux4567

Dec 01, 2014
4:19 PM EDT
I see the Red Hat spokesperson in disguise on lxer (me1010) now even needs back up from a Red Hat employee (hughesjr from Centos, which is now owned by Red Hat).

Red Hat seems to be very concerned to avoid a huge image disaster with systemd, but it's way to late for that at this point.

I cannot believe Red Hat seriously thought to get away with such a bully behavior (as seen with systemd) and still hope to be considered the poster child for a well behaved open source company.
Ridcully

Dec 01, 2014
6:00 PM EDT
I find it amazing that an uncomplicated report on what I have observed is instantly attacked by the "pro-systemd in-group". Apparently, my simple statement of facts can't be attacked and instead, I have been attacked via the more subtle mode of "psychology" and "confirmation bias". Incredible is the only word I would use, to describe the lengths the "systemd in-group" will go to in order to discredit an opposing viewpoint.

Let me put it more bluntly, me1010. I think that Red Hat's moves are now wide open to view and finally being understood for the distasteful and underhand operations that they are. What is even more disturbing, is that this is coming from a very successful Linux company that (supposedly) understands Unix/FOSS principles and has no need to resort to "Redmond-style tactics".

I repeat, I am not a programmer or developer, but I can read, and I can see what is happening. I do not believe "comfirmation bias" is involved, purely because I had none to begin with and simply watched the way the pieces fell into place and finally stated that more and more people were confirming my original position from the earlier days of the turmoil. Until this fracas began, I had never heard of either sysVinit or systemd......BUT.......I did know the Unix philosophy and thereby, the development philosophy of Linux. Given what has happened (and especially the move to fork Debian by the Unix developers), I find it hard to believe, me1010, that you can still sit back and say that systemd is in full compliance with the Unix philosophy and, as linux4567 notes above, that Red Hat seriously thought to get away with such bullying behaviour in a Linux world. And there, I think are the real problems.
Ridcully

Dec 01, 2014
6:15 PM EDT
Let me hasten to add, me1010, that when I used the words "you can still sit back and say that.......", I was NOT speaking to YOU personally, ......I was using the general form meaning "anybody".....but given your postings, you might also be included in the "anybody" ? It's your call and I make no unfair assumptions.
me1010

Dec 01, 2014
6:31 PM EDT
@Ridcully:

Here's a test.

systemd: who owns the copyright? Is it:

A. Poettering and other individual developers.

B. Red Hat

C. Microsoft

D. Al Gore

E. Ridcully

F. me1010

?

What would happen if:

Poettering and friends decide that Red Hat doesn't pay enough, and so they get Intel to pay their salaries instead?

A. systemd continues to be a RH 'pushed' product.

B. Poettering and friends continue developing systemd.

C. Ridcully decides systemd is now an Intel power grab.

D. systemd will still be used by most GNU/Linux distros.

Choose all that apply.
Ridcully

Dec 01, 2014
6:47 PM EDT
@me1010........I find the above posting a more or less ridiculous exaggeration for emotional effect. Ignoring it almost completely, my observations suggest that while systemd may be opensource and therefore subject to the GPL and in that sense, theoretically everybody "owns" it, the practical side is far more unpleasant. Bluntly and practically, Red Hat "owns" it and is pushing it via Poettering. Finally, Red Hat most definitely IS using their enormous community power to force systemd into every other distribution. The Debian fork is the first positive and practical rebellion against the "Red Hat Empire". I believe I have answered as much as needs answering in the above post.
me1010

Dec 01, 2014
7:05 PM EDT
@Ridcully:

wrong.

RH does not own systemd. Poettering does.

Copyright is ownership. GPL code does not belong to everyone. Please see recent practical applications of copyright and GPL versus public domain at github...

https://github.com/blog/1530-choosing-an-open-source-license

It is a common misconception that GPL code is interchangeable with public domain code. But this is false. GPL code is publicly released, useable, changeable, but not publicly owned. The end-user is granted a license to use and change the code, but this does not imply public ownership.

See the GPL creator's thoughts -- queue RMS:

https://www.fsf.org/blogs/rms/public-domain-manifesto
JaseP

Dec 01, 2014
7:13 PM EDT
Another thing needs to be brought up, which was brought up in the comments to the original article,... Calls to fork distributions to be maintained without systemd, by systemd proponents are disingenuous. Many have stated that even the Debian developers themselves are incapable of maintaining the project without the systemd dependencies. They just do not have the time/resources to do so (which is why, ultimately, I feel that Devuan will fail,... unfortunately).

Yes,... it' open source,... so you are free to fork, adapt the code, etc. However, saying that when it is obviously not feasible is facetious. Even FreeBSD will have to import some measure of systemd code (logind and dbus/kdbus dependencies) to maintain the availability of various DEs (GNOME, KDE, etc.). We are left with the awkward situation where the least dependency on systemd will be had with things like Android and Chrome.

Ultimately, the only options for people who don't want to play RedHat's ball is that they are either going to have to jump ship and fork/patch everything they like using that has a dependency issue (Devuan or switching to FreeBSD for example),... Or they are going to have to jump in and patch/code and alternative to systemd to remove the "nasty bits"...
kikinovak

Dec 01, 2014
7:35 PM EDT
The "choice" presented by systemd proponents reminds me of the famous Henry Ford quote. "Any customer can have a car painted any colour that he wants so long as it is black."
JaseP

Dec 01, 2014
7:36 PM EDT
@kikinovak:

You said it!
Ridcully

Dec 01, 2014
8:12 PM EDT
@me1010.....Please stop splitting hairs. Pedantically, I suppose you are correct about the GPL. As far as I am concerned, the GPL means that anybody can use and modify the code, provided that, when they release the code under their modifications, then all of that modified code must be under the GPL as well. Okay, that to one side, I now deal with something else you indicate.

If what you are saying is correct in that Poettering OWNS systemd, then do I assume it is his under a copyright, that it is NOT released under the GPL and that he has ultimate say over who may use it and do anything to it ? IF that is the case, then the situation is even more dangerous than I first believed. Or are you saying that systemd IS released under the GPL and that Poettering is the author and in that limited sense only, "owns it" ?

Either way, I stand by what I stated above. Red Hat is using its enormous community leverage as a large Linux organisation to force systemd onto the rest of the Linux community. And I do not resile from using the word "FORCE".....it is the only descriptor that fits what is happening. Kikinovak has also summed up this situation perfectly. Red Hat is creating a situation where you can use any flavour of Linux you like, just as long as you use Poettering/Red Hat's systemd......after all, it is Red Hat that is forcing this - with Poettering's cooperation of course.
me1010

Dec 01, 2014
8:25 PM EDT
@Ridcully:

Poettering is the copyright owner. I believe there are a few components copyrighted by other individuals. Code released under the GPL requires a copyright notice. So your statement regarding GPL versus copyright owner doesn't follow the logic or the rules of the GPL.

I'm not splitting hairs here. It's highly important that people understand the difference between public ownership via public domain and licensed freedom via the GPL. If there was no difference or if it was just 'splitting hairs', why all the fuss over the GPL? Why wouldn't RMS just release GNU software under the public domain and be done with it? The reason is ownership and control. The GPL allows authors to remain as authors and to hold control over the use of publicly 'free' code. Without the GPL or other open source licenses [vs just public domain], it's very likely we wouldn't have anything to argue over here -- since GNU/Linux wouldn't exist...

RH is not forcing anything, nor does it have the power to do so. Please provide proof... so far no one has provided proof, just empty and mostly nonsensical allegations.
linux4567

Dec 01, 2014
8:37 PM EDT
@Ridicully, you are right and me1010 is wrong. With the GPL it doesn't matter who formally owns the copyright, because the copyright owner gives away practically all rights to his code, except the right to sue anybody who changes the license when making derivative works or who doesn't provide (at least on request) the source code to the users of the binaries. That's why the GPL is also called the 'copyleft' license.

@me1010: your distraction tactic is a waste of time, nobody will fall for your hair-splitting with regards to the GPL and it's very transparent that this is simply an attempt by you to derail the thread by focusing on non-issues of no relevance to the subject of the thread.
me1010

Dec 01, 2014
8:55 PM EDT
@linux4567:

No. It does matter who the copyright owner is... and this is one reason Debian did not choose Upstart -- the copyright assignment required by Canonical.

The copyright owner may change the license of the software, and many do... especially the smaller startup ISVs who release code as FOSS to get started. Copyright ownership matters. And if Red Hat actually wanted to 'control everything via systemd' they would require copyright assignment.

EDIT:

See here: https://wiki.debian.org/Debate/initsystem/upstart

Ridcully

Dec 01, 2014
9:13 PM EDT
Okay.....we now get to the point where we know that systemd is issued under the GPL. Poettering wrote the software and holds the copyright, but that copyright is subject to the prescriptions of the GPL and therefore the source code of systemd is open to anybody to modify without Poettering's approval, provided that they then release the modified systemd under the GPL. Fine. But we had to fight to get to this point and it's both distracting and time wasting in that it was suggested that Poettering "owned" the code without stating the rest of the facts as they applied to the GPL.

@me1010.......I disagree with you completely over the forcing of systemd onto the Linux community by Red Hat. The proof is in what you see happening - Red Hat is implicitly forcing the code by its community leverage and by the fact that it is now being added and abetted by its major DM, Gnome. Of course there is no statement out of Red Hat (or Poettering) saying "Linux community, you WILL use systemd, or we have ways of making it so darn hot for you with broken routines, that you will be very sorry you ever tried to stop us.".....but the practical effect of what they are doing is just the same. systemd is steadily encroaching on other initialising routines (and that is well documented by others) and if you don't use systemd, you can be in big trouble and be forced to rewrite considerable software. The biggest mistake was for Debian to move and I understand there were successive challenges to that decision and eventually culminated in the fork.

Thanks linux4567.......I sometimes wonder if I am barking up the wrong tree, but it's very nice to know others think the same way too.
me1010

Dec 01, 2014
9:25 PM EDT
@Ridcully:

Correlation does equal causation. Simply because non-RH distros are adopting systemd does not prove that RH is controlling the choice of init via systemd. You've placed intention where there is none.

The 'time wasted' on GPL, copyright, and licensing is not wasted at all. It should be very clear to those who may be following the argument that copyright implies control -- and thus the copyright owner is the controller -- and in systemd's case that controller is not Red Hat.

Forked GPL code still requires copyright statements. You don't need a separate permission, as permission is granted in the GPL. However, the copyright owner may still decide to change licenses for all future releases. So, in a very real way, if you use FOSS you are always at the mercy of the copyright owner whomever that may be -- unless you are very good programmer and keep your own software stack.

EDIT:

Actually I was wrong about Poettering being the copyright owner. He is the owner of a significant portion of the code, however -- Intel does in fact have its stamp on several header files. And there are other authors too... However, there are very few Red Hat copyright notices within the source tree. So, actually the statement: Intel is pushing systemd. is more correct than: RH is pushing systemd. However, neither is truly accurate.

jdixon

Dec 01, 2014
9:40 PM EDT
> Linux is open source.. So fork it and do what you want. Or use the fork these guys create. Then we can see who wins in the end.

Oh, I will. The one thing that these discussions have convinced me of is that I want nothing to do with systemd. If necessary I'll be looking at the BSD's. Systemd is far too reminiscent of early KDE4 for my taste.
Ridcully

Dec 01, 2014
9:43 PM EDT
@me1010........The distros are adopting systemd because they have to......are compelled to......I cannot make it clearer. We are wasting time here. You are playing semantics. By employing Poettering, encouraging him in his work, by seeing Gnome fall into line given that Gnome is Red Hat/Fedora's default DM, Red Hat is implicitly cooperating in seeing that systemd is adopted as widely as possible. I direct you once again my to my statement above in inverted commas which says:

Of course there is no statement out of Red Hat (or Poettering) saying "Linux community, you WILL use systemd, or we have ways of making it so darn hot for you with broken routines, that you will be very sorry you ever tried to stop us."

Also as my final comment, no users or systems administrators have ever been this upset before in my experience and their concerns are already widely documented. I don't have to repeat them. The mercy of the copyright owner under the GPL is a theoretical one only; it has no practical value provided the programmer obeys the strictures of the GPL, and if you as a programmer, pick up GPL'd code to modify, you automatically know you must obey the GPL rules - or pay the penalty....again, your writing is a red herring.

I don't think I need to continue this debate. I have said all I need to say except that systemd is now an unhappy blot on the overall structures of Linux - but Red Hat is happy, and is in control, and that's really what this is all about.
me1010

Dec 01, 2014
10:11 PM EDT
@Ridcully:

OK... you have your opinion on systemd, real or imagined, and there is no way to convince you otherwise.

However, you've missed the mark on the GPL. The power of the GPL is the ability to give source code without giving up authorship. It is not a theoretical point. The following example happens all the time... A given program is released as GPL... it builds a market share... the copyright owner switches from the GPL to a commercial license. The more common method -- because end-users may not like the bait and switch -- of extracting cash from GPL or other FOSS-ware is through proprietary add-ons. Similar to RibbonSoft's QCAD's DWG add-on [damn them to hell for that!] ...

Ridcully

Dec 01, 2014
10:55 PM EDT
Whoa back, me1010........if something is released under the GPL, it is irrevocable as far as I am aware. It is possible to have add-ons which are personal property and not part of the original GPL'd release, but my understanding of the GPL is that once software is released under that licence, that is where it stays........it cannot be taken back. Here's the reference:

http://www.rosenlaw.com/pdf-files/Rosen_Ch06.pdf
me1010

Dec 01, 2014
11:24 PM EDT
@Ridcully:

You are thinking of the end-user... or the licensee. My statements refer to the author/s. And the authors listed in the copyright statement of GPL code are free to choose any new license he/she/they wish for any future code based on the GPL code he/she/they released.

It is true that in many cases there are many authors, and permission must be given by all authors in order to change the license. This is why Canonical contributors complain about the requirement for copyright assignment. If you contribute GPL code to a Canonical copyrighted project, you are required to forfeit your say [yea or nay] in changing the license from GPL to commercial. This is one reason I do not use Ubuntu.

Ref: I'm sure there's a better reference... but I'm too lazy at the moment:

http://stackoverflow.com/questions/9116172/gpl-and-lgpl-to-commercial

maybe wikipedia too...

https://en.wikipedia.org/wiki/Business_models_for_open-source_software#Re-licensing_under_a_proprietary_license

Meaning: Your own software can be re-licensed proprietary and also then combined with other people's FOSS if that other FOSS uses a 'permissive' open source license.
mrider

Dec 02, 2014
12:56 AM EDT
A person that holds a copyright can revoke or change the license at any time. That includes GPL. The trick is that the copyright holder cannot retroactively revoke the license on the code already released. That means that if the users of the software do not agree with the change in license, they can always fork the code at that point and continue onward with the code as it was at the last GPL release.

A perfect case in point is MySQL versus MariaDB. Now that Oracle owns the copyright on MySQL, they can license it however they wish. But, MariaDB is free to continue onward with the code as it stood just before Oracle bought the copyright. However, MariaDB cannot change the license because they do not own the copyright. They can however dual license their own code - I have no idea if they do that.
Ridcully

Dec 02, 2014
1:08 AM EDT
THANKYOU mrider......what you have written is what I intended to convey in my post above, but apparently didn't explain it properly. Your word "retroactively" fully deals with what I meant by saying that once code is released under the GPL, that code cannot be changed from the GPL.....Your example is perfect.
Fettoosh

Dec 02, 2014
2:10 AM EDT
me1010 is saying the same thing when he said future code meaning release
Quoting:And the authors listed in the copyright statement of GPL code are free to choose any new license he/she/they wish for any future code based on the GPL code he/she/they released.
Ridcully

Dec 02, 2014
2:38 AM EDT
Yes, he is Fettoosh, but that occurred in a posting after his original post to which I offered an objection. Ultimately, all three of us have the same concept I think......well, I am sure mrider and myself agree.
me1010

Dec 02, 2014
10:04 AM EDT
Yes... it would appear that we all understand the GPL after all...

However, it is important to note that it may be successfully argued that:

1
int main
{
int x=0;

x++;

return 0; }


2
int main
{
int x=0;

x++; __asm__ ( "nop;");

return 0; }


:may constitute enough of a change that code 2 is a "new" release and can be re-licensed as proprietary.

What I'm trying to show is that the end-user of GPL-ed code is certainly free to use the original version of the code in a 'free' way. However, any and all future releases and updates - including security updates - [made by the copyright holders] are not required to be released under this same license. And so, as I indicated before:
Quoting:in a very real way, if you use FOSS you are always at the mercy of the copyright owner whomever that may be -- unless you are very good programmer and keep your own software stack.


The fact that RH does not control the systemd copyrights makes the entire argument that:
Quoting:RH is controlling everything by pushing systemd
entirely specious and/or universally applied to all FOSS.

However there remains the following:

1. systemd code quality is poor.

1a. Measurable, but no one here has provided a yard stick.

2. systemd is philosophically flawed.

2a. OK.. can't really argue that one -- but it is a personal opinion and not a factual statement.

3. systemd is prone to scope creep.

3a. The main inclusion of code has been udev... which has already been forked.

3b. The libc choosen may be a problem for some use cases, but there are solutions 'out there' to build anyway.

4. systemd's maintainers are sometimes rude or otherwise unpalatable.

4a. OK... again, this is a personal opinion and is at least partially driven by the non-systemd party... as my parents are fond of saying: It takes two to tango.

So... the entire argument over systemd hinges on very few factually based decision points. And hopefully, the RH conspiracy theorists are beginning to rethink the soundness of that wild accusation.
Fettoosh

Dec 02, 2014
12:18 PM EDT
OK, so far, according to few posts here, two main programs, udev & D-bus, were included in systemd that consequently made many applications that use or depend on these two programs also dependent on systemd.

Now udev, according to Wikipedia, was developed by Greg Kroah-Hartman and Kay Sievers.

& D-Bus, again according to Wikipedia, was developed by Red Hat and the community developed as part of the freedesktop.org project.

Quoting:Heavily influenced by the DCOP system used by versions 2 and 3 of KDE, D-Bus has replaced DCOP in the KDE 4 release. An implementation of D-Bus supports most POSIX operating systems, and a port for Windows exists. It is used by Qt 4 and GNOME. In GNOME it has gradually replaced most parts of the earlier Bonobo mechanism. It is also used by Xfce.


The questions I have, and I appreciate some answers are:

- Were these two programs forked/hijacked or were they voluntarily include into systemd by all their developers?

- was it really technically necessary to do the inclusion? what are the reasons? Did RH address or justified these concerns?

- how big of an effort is it to maintain a fork to keep them separate by others than their original developers?

- Has any one objectively addressed such concerns?

- Has RMS, or any one of the main FOSS advocates made his/her opinion about this issue public?

- ...
me1010

Dec 02, 2014
12:28 PM EDT
@Fettoosh:

eudev is the fork of udev:

http://www.gentoo.org/proj/en/eudev/

I don't believe D-Bus code is included in the source tree of systemd. My understanding is that systemd uses D-Bus as its messaging transport. So systemd depends on D-Bus, rather than D-Bus depends on systemd... please correct me, if I am wrong... but please do not post third party hearsay as proof.

If you want to know the answers to the rest of the 'why' questions, check out Poettering's blog -- whether anyone likes him or not, you will find the 'why' questions mostly answered there...

http://0pointer.net/blog/
Fettoosh

Dec 02, 2014
1:52 PM EDT
Quoting:but please do not post third party hearsay as proof.


That wasn't my intention. I was merely assuming the worst as portrayed by some posts .

In my opinion, if Poettering is heavily involved in this project, his blog might not be a reliable source for facts concerning this subject. If you have one, I would rather read an independent unbiased source instead.

If you notice, my questions are aimed at the claims in the various threads hoping someone does real evaluation and state the facts instead of interpretations & assumptions.

gus3

Dec 02, 2014
2:28 PM EDT
An example of confirmation bias:

A psychic tells you, "You will converse with a drop-dead gorgeous blonde today..." And then it happens. Wow, the psychic was right? Only because you hoped so.

And a counter-example, same psychic:

"...right before you get hit by a truck." And then it happens. Ummmm...
me1010

Dec 02, 2014
3:07 PM EDT
@Fettoosh:

I don't have a blog, I'm a terrible blogger. I don't considered myself unbiased, because we are all biased - although I do try to remain unbiased.

Arch has really good documentation.

https://www.archlinux.org/packages/core/x86_64/systemd/

I have referred to Arch's documentation in other threads. I really should start using the distro, but I sort of 'grew up' with Debian. Here's the build dependecies they list as well as which programs currently employ systemd hooks. I'm sure many of the anti-systemd people will say:
Quoting:Hey! See! That's 133 programs that require systemd.


However, this is not true, most of those 133 programs include an 'if it's there' optional behavior -- sort of like [or maybe is] make's ifdef.

For examples CUPS:

grep -r systemd *


config.h.in: * Do we have systemd support?
config-scripts/cups-systemd.m4:dnl   systemd stuff for CUPS.
config-scripts/cups-systemd.m4:dnl Find whether systemd is available
config-scripts/cups-systemd.m4:AC_ARG_WITH([systemdsystemunitdir],
config-scripts/cups-systemd.m4:        AS_HELP_STRING([--with-systemdsystemunitdir=DIR], [Directory for systemd service files]),
config-scripts/cups-systemd.m4:        [], [with_systemdsystemunitdir=$($PKGCONFIG --variable=systemdsystemunitdir systemd)])
config-scripts/cups-systemd.m4:if test "x$with_systemdsystemunitdir" != xno; then
config-scripts/cups-systemd.m4:        AC_MSG_CHECKING(for libsystemd-daemon)
config-scripts/cups-systemd.m4:        AC_SUBST([systemdsystemunitdir], [$with_systemdsystemunitdir])
config-scripts/cups-systemd.m4:        if $PKGCONFIG --exists libsystemd-daemon; then
config-scripts/cups-systemd.m4:        	SDCFLAGS=`$PKGCONFIG --cflags libsystemd-daemon`
config-scripts/cups-systemd.m4:        	SDLIBS=`$PKGCONFIG --libs libsystemd-daemon`
config-scripts/cups-systemd.m4:if test -n "$with_systemdsystemunitdir" -a "x$with_systemdsystemunitdir" != xno ; then
configure.in:sinclude(config-scripts/cups-systemd.m4)
debian/cups-daemon.install:lib/systemd/system/cups.socket
debian/cups-daemon.install:lib/systemd/system/cups.path
debian/cups-daemon.install:lib/systemd/system/cups.service
debian/libcups2.README.Debian:Socket-activation under systemd

debian/libcups2.README.Debian:Since 1.7.1-7, when systemd is the active init system, CUPS can be

debian/libcups2.README.Debian:its sockets. The socket definition in /lib/systemd/system/cups.socket gets debian/libcups2.README.Debian:configured through files in /etc/systemd/system/cups.socket.d/; see debian/libcups2.README.Debian:systemd.unit(5) and the examples files in /usr/share/doc/cups-daemon/examples . debian/libcups2.README.Debian:/etc/cups/cupsd.conf. For systemd's socket activation to working, cups.socket debian/libcups2.README.Debian:configuration files in the /etc/systemd/system/cups.socket.d/ directory.


I separated out the important point.

As far as code complexity is concerned, it's not an easy topic of research, especially considering systemd. In general there are many measures of code maintainability, cost, and quality. However, SLOC [source lines of code] is a key value in the calculations of code cost and maintainability. Based on SLOC alone, systemd could probably be calculated to be cheaper to maintain than a a full laden general purpose desktop deployment of sysvinit and associated smaller programs --- although I haven't done the actual calculation.

Here's a wiki link to source code quality measurement -- much of it will seem esoteric:

https://en.wikipedia.org/wiki/Software_quality

At one point in my meandering career, I took a few nearly forgotten courses on CMMI:

http://whatis.cmmiinstitute.com/grow-your-career

But I found it very boring, I much prefer to put things together with my hands than do esoteric calculations in risk management. I suppose that's why risk management pays lots.

@gus3:

No harm will come to me as the bearer of the holy hand grenade of Antioch.

EDIT: corrected 144 to 133.
theBeez

Dec 03, 2014
2:17 PM EDT
@me1010, look here: mockus.us/papers/decay.pdf There's nothing esoteric here. It's a statistical analysis which code goes bad. There is a significant overlap with systemd.

Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]

Becoming a member of LXer is easy and free. Join Us!