レナート   Wunschkonzert, Ponyhof und Abenteuerspielplatz   ﻟﻴﻨﺎﺭﺕ

Posted by Zeeshan Ali (Khattak) at Sat Nov 13 22:25:18 2010
Sounds like one of those "I am way too smart to write code so I'll just laugh/criticize other's work" dudes out there.

Posted by Jeremy at Sat Nov 13 23:09:39 2010
Cry more!

Posted by elmarco at Sat Nov 13 23:27:22 2010
Damn, you had a twin you didn't know!?

Posted by loupgaroublond at Sat Nov 13 23:30:29 2010
Hey Lennart,

I don't know who the presenter is, so i can't vouch for any real coding abilities, but i can say there are people out there with a radically different vision of the Free/Open desktop who are in fact coding it. I can't say i agree totally with them either, but there is a point of view that's valid and not FUD. I've come across it often in hacker circles, and it's definitely interesting to see who says it.

Chances are, this is probably going to be one big trolling session. Still, there is such a point of view out there.

Posted by Jonatas Esteves at Sat Nov 13 23:51:57 2010
Only needed to read down to ""Yo Dawg!" Stacking layers of redundancy. (Phonon -> GStreamer -> Pulseaudio)" to know there's some ignorance in this paper. Can't speak for the entire publication, though, as I didn't read it all.

Posted by wonderer at Sat Nov 13 23:59:35 2010
Hy,

just on the two "questionmarks" you set up. Yes, its one essential of the congress that you can be anonymous if you want and its much alike acceptable for all CCC members.
But you are free to attend to the event or ask for the guy's mail so you can reach him and help him to improve his talk!

Posted by jake at Sun Nov 14 00:49:26 2010
This talk looks interesting. It might be nonsense, but it might be good. The summary seems informed enough. Certainly the stack Phonon -> GStreamer -> Pulseaudio contains redundancy as does the whole of the  free desktop. This is a result of the freedom, but nevertheless it is worthwhile to do some refactoring now and then.

I partly agree with elmarco :-)

Posted by Anders at Sun Nov 14 09:58:02 2010
Your post sounds like FUD-FUD to me. How about explaining what he got wrong?

I think he raises some valid points, but I'm not a GNOMEr, so the chances are high that I missed something ;)

Posted by Alexey at Sun Nov 14 10:36:42 2010
I also agree with many points in the submission. Been spending most of my linux maintainance time on trying to fix the issues coming from the listed software (dbus, polkit, consolekit, hal, networkmanager, gdm, etc).
luckily, pulseaudio is something that can be completely avoided, and i'm 100% happy with what ALSA does.
and i'm also free software developer.

Posted by Rui Matos at Sun Nov 14 14:20:49 2010
Changing the status quo is hard. Thanks for doing the hard stuff Lennart!

Posted by Samium Gromoff at Sun Nov 14 14:53:18 2010
Structural elegance is hard, and FOSS desktop fails at it, IMO.

Mind you, I respect the work you did and I'm grateful that you've done it, but there is a slight problem with your adamant rejection of criticism.

Posted by Alex at Sun Nov 14 15:51:42 2010
I wonder if we could get someone high-profile like Mark Shuttleworth to make a post explaining the difference between Phonon, GStreamer, and Phonon.  Maybe then people will listen and understand.  To everyone else commenting on this post: make sure you know what you're talking about before you start running your mouth off.  Some of you clearly don't know anything about the technologies mentioned above beyond "it does something with audio".

Posted by Rupert at Sun Nov 14 19:04:29 2010
I afraid I don't see any FUD in that. Several highly valid points, though. But what would I know, I just maintain hundreds of Linux systems for a living.

Posted by Diego at Sun Nov 14 21:26:23 2010
It's called "paper trolling". I mean, isn't he suggesting that we use X11 as a generic IPC system? We shouldn't be using it even for graphics!

Posted by me at Sun Nov 14 23:23:23 2010
it's a "conservative".

Posted by nona at Mon Nov 15 00:41:50 2010
Well, irrespective of whether this talk ends up as FUD or not (the talk may end up more sane than the somewhat provocative summary suggests), maybe there will be some valid criticism in there.

I like all the projects he seems to have issues with, I believe they advance the state of the art of the Linux Desktop, but I don't expect them to be perfect. It's normal to discover flaws in the design or implementation when the rubber meets the road. I thought HAL and upstart were an improvement over what we had historically, but that didn't mean that we couldn't do even better with udisks/upower and systemd.

So I don't think we should resist a (maybe?) FUDtalk like this, because even if only 10% of the talk is valid criticism, it can only expose weaknesses which we can then fix.

Posted by tante at Mon Nov 15 09:18:14 2010
To be honest, it sounds a lot like the talk Fefe gave last year about "bad APIs", so somebody probably just copied his style to rant some more. Just trolling.

Posted by esarbe at Mon Nov 15 11:59:35 2010
Uh... he's criticising that people learn from mistakes (HAL), comparing apples with pears (Phonon, GStreamer, pulseaudio), disdains auto-setup framework and tools and seems generally not up-to-date with his knowledge (dbus network transparency).

FUD.

Posted by seb at Tue Nov 16 12:06:04 2010
calm down, Lennart. Some notes:

1. The summary (and maybe the whole talk) being provocative doesn't mean it's all wrong.

2. The speaker may not be listed, because the "Fahrplan" is not yet completely done.

3. Many people view multiple abstraction layers, functional duplication, automatic configuration, registries, large package dependency chains, etc. as a problem. Let's hear them out.

Posted by Vincent at Thu Nov 18 18:31:54 2010
Too bad he is mostly right. I'm a sysadmin and can tell you that some things are horrible with all these *kits. I tried to disable hibernation for a group of users. However policykit documentation for sysadmin is nonexistant. If something is documented, you can be pretty sure that it's already obsolete. Also, distributions (ubuntu here) like to place various config files in some non-obvious directories. Is there any book, or up-to-date documentation that describes ConsoleKit, PolicyKit, Udisks, DeviceKit for admins? I'm just a sysadmin, I don't have time to read source code.

Posted by datenwolf at Fri Nov 19 21:20:30 2010
- I don't know why my name is not appearing in the event details, you've to ask the organizers, why they didn't include it into the schedule.

- The talk is subtitled "The Good, The Bad and the Ugly". I'll point out IMHO very good decisions on the API as well.

- My thoughts on the topics outlined have two origins:
--- ATM I'm developing my own experimental desktop environment, in which I want to explore new usage and interface patterns. While this DE will be very different from KDE, GNOME, XFCE and the like, I nevertheless develop it alongside the Freedesktop standards.
--- I'm a systems administrator in a University's network, with a 3.5k userbase. Distributions like Ubuntu or Fedora no longer work "out-of-the-box" due to all those "make things for the user easy" automatisms which hilariously clash with our needs. Un-configuring all these for every new release of the distribution(s) is annoying like hell — it's especially ConsoleKit causing the problems here.

- Of course I waned to provoke with the abstract. And I want people who consider attending the talk to look up those terms beforehand, and look into their system. I bet, that most people don't have a clue, what's running on their systems without noticing.

- The Stack Phonon->GStreamer->PulseAudio is not meant as criticism of the single blocks themself. I'm critisizing, how these things are connected and how and what for most programs use it.

- PulseAudio is something I really like.

- D-Bus is network transparent in theory (tcp and nonce-tcp transport). But did you every try it? If something requires 3 days of tinkering to make it work, then it's not transparent.

- The X11 vs. D-Bus IPC topic focuses on a very limited set of IPC operations for which D-Bus can be used for but adds additional complexity with only very little benefit above traditional X11 IPC.

- And I'll point out an actual case of a security nightmare caused by the interaction between ConsoleKit/PolicyKit (namely the PAM configuration shipping with Ubuntu >9.04 to make them work) when they got integrated into a preexisting Kerberos-5 infrastructure. Technically it was a misconfiguration, but it went by undetected, since the testsuite didn't account for the kind of hole it opened. The case here is, that PolicyKit and ConsoleKit are extreme poorly documented and at that time noone in the admin group had a clue what was going on.

Posted by Brent at Tue Dec 28 23:42:21 2010
TROLOLOL

Posted by Lennart at Thu Dec 30 11:29:01 2010
datenwolf: the dbus session bus is not supposed to be shared across the network. And the dbus system bus even less, hence it doesn't matter whether dbus is encrypted or not, since the busses are only accessed locally. X11 is not an IPC layer, it is in fact completely unsuitable for general client-to-client communicaion. In some cases it has been tried to use it for IPC, but read stuff like the XSETTINGS spec as an example how awful that actually is.

It's uncool doing talks whose only contents is bitching about other people's code. It's even worse being badly prepared and having no clue of the stuff one is talking about when doing that.

Posted by Erik at Thu Dec 30 21:06:03 2010
Was it really necessary to be that offensive at the conference? I mean everyone could see that Datenwolf was not right on many specific points. And it's ok to tell him that. But starting a comment with "Do you hate handicapped people?" is not cool. He obviously hadn't thought of that when he was criticising that Gnome starts a full session for a login.

Posted by Dominik at Fri Dec 31 02:06:27 2010
One short quote from Tim Pritlove:
"...if there not really delivering good information or delivering it in the right way they could (the speaker) face a angry mob in the audience criticizing every detail of what they said so they expect a very engaging strong community..."

Maybe the speaker has this time not really got good information to deliver and so faced the mentioned mob.

I think it was entertaining and I even learned something. So thank you lennart!

Posted by stfuorstayhome at Fri Dec 31 18:15:16 2010
there was no mob. there was just lennart.

Posted by Armin Ronacher at Fri Dec 31 18:18:04 2010
> He obviously hadn't thought of that when he was criticising that Gnome starts a full session for a login.

They why did he have to bitch about it in the first place? There are reasons why systems are complex.

I applaud Lennart for what he did there and what he does for Linux on the desktop in general.

Posted by Pyr at Fri Dec 31 19:47:28 2010
It's awful to see how comments here are censored. I made a previous comment, there was also a comment by a guy named "nert" that was posted here earlier but removed later, obviously.

This clear demonstrates a lack of self-criticism on the side of lennart. Just like it could be seen during the talk: as soon as your work is criticized, you try to take the word away from the speaker.

@Armin: I see the duty of showing what specific systems are there for on the side of the people who write these systems. The documentation of many, many projects that lennart is involved in can be described as "lacking" at best. If nobody understands a system without specifically asking the creators of said system, something is wrong. The "why didn't you talk to me first?"-mentality that lennart exhibited during the talk demonstrated this perfectly.

People need to understand that open source software is not really much better than closed source software just because the code is released. Without documentation (not just comments in the code, but clear explanation of design decisions, too!), 98% of the good points of OSS are void.

Posted by cedric at Sat Jan 1 09:00:41 2011
If the admin giving the talk pretend that many problems he encounter from linux desktops now tend to come from automatic complex software that does not work, well I do not see why his views should be discarded, especially since so many other tech people feel the same.

Posted by Lennart at Sat Jan 1 16:09:37 2011
Pyr: well, if somebody call me names I will delete these comments. I don't think I need to offer a platform here for people to insult me. That I handle these things this way is written explicitly on the bottom of each page. If you want to insult me, then you may do so on your own blog.

Pyr, also, you should be aware that the docs for most of my projects are actually relatively good, and in fact people who actually bother to read them do acknowledge that. For example here: http://thread.gmane.org/gmane.linux.redhat.fedora.devel/144124 Hence, Pyr, you are fudding around. Also, if Wolfgang wants to do a talk about something he should do his homework and actually acquire the necessary information and knowledge to be able to do a good talk. He didn't do it. But that doesn't enable him our you to claim that that was my fault, and I should have informed him. If he does the talk it is his duty to prepare himself, not mine.

Posted by Markus at Sun Jan 2 03:36:50 2011
I've seen the talk via the wire, if you missed it, there is a low quality version on youtube:
http://www.youtube.com/watch?v=ZTdUmlGxVo0

Given the datenwolf saw the post in this blog, participated in the comments, he really had a chance to sent his slides for review, peer review exists.

The presentation was not informative in any way due too lack of background, and it is really hard to make a point about a system if you lack the background, this is similar to the church claiming the world is flat. Like the church, the presenter could have known better, the knowledge already existed, it was just ignored.

I was totally glad there was somebody with the required background in the audience, who was willing to stand up on the claims made.
That made the difference.

I found this thread as I was looking for some recent news on systemd, so I definitely got no personal fixation for Lennart, and I even disagree with some (optional) concepts of systemd, but if it is optional, who cares.
Independent whether you like the outcome of what he does, one has to admit he is pushing things, he makes a difference. Even if everything he does would end in the trashcan for unusable software, others could learn from it.
I'm surprised myself why a simple desktop on a 64bit machine needs that much memory, but either I can simply buy more memory, write patches to reduce useage or, which is of course the cheapest option, bitch about it, which does not change anything, but gives me the opportunity to ... waste my audiences time?

That said, I'd appreciate a talk about "'recent' changes on the linux desktop" given by Lennart on the 28c3, including things like predecessors and documentation and loc involved, so the ladies and gentleman who missed the last 20 years of development get a cheap possibility to get synced with reality.

Obviously he'd been a better choice for the topic anyway.

Overall the presentation by the datenwolf was a waste of time, the only way it could have been worse was without somebody questioning his claims.


One has to give the datenwolf some credits though.
Continuously picking fights with the maintainer/designer/friendoftheauthor/... of the software he criticized in front of a public audience while loosing each round is definitely noteworthy: I'd he was by far the most ignorant presenter on the whole conference giving the worst presentation.

That said the criticism in the initial blog post was correct and could not have been more correct.

Posted by Hans at Sun Jan 2 04:23:00 2011
http://sumosu.soup.io/post/98484486/Linux-Bully-Beatdown-at-27C3-feat-Datenwolf

Sorry but that guy Datenwolf is one of the biggest Noobs i ever seen on the ccc congress

Datenwolf you should bitching about your english accent

fail

Posted by nn at Sun Jan 2 05:16:44 2011
I think it's fair to criticize anything that is unsettling about administering Linux. I've had my own battles with destkop administration for a visual effects studio with both Linux and OSX. It's quite unpleasant, but amazingly enough, Apple listens to some extent, but it seems any criticism falls on deaf ears when it comes to Linux.

I think there's an apparent problem with heavy commercialisation of Linux that's leading to a divide by the development community which produce free software, that companies are leveraging. But at the same time, companies are trying to design the ideal profile, the ideal use-cases, the one-size fits all mentality and failing pretty hard.

This shows, because it's definitely not the desktop market making profits for the companies that sell Linux services and have a responsibility to shareholders: Redhat, Novell.

I think datenwolf has valid frustrations, perhaps he tried to be vocal about with the community, but failed to email the "right person", because apparently mailing lists aren't the place anymore to discuss issues. But there's another hinderence, either datenwolf and his crew needs to hire more people (i.e. spend more money) to develop a very customized version of Linux that fits their heterogeneous setup, or find a compromise with the distributions that he has to work with. I think it ends up being the latter for most admins.

Anyway, I just want to say there's definitely a clear disconnect between developers, vendors, and end-users. It seems the reason why noone criticizes projects directly anymore is because it's filtered by the communities surrounding the products.

I think Lennart tried to oversimplify the situation: "don't like it? don't use it!"

How about: "don't like it? why are you a sys admin? why don't you just shoot yourself now and let us developers get on with our shit."

It just doesn't help.

Posted by Tobias at Sun Jan 2 18:24:25 2011
Seen the video.

The whole thing is cringeworthy. And both sides were embarrassing themselves in their own peculiar ways.

Posted by Infact at Mon Jan 3 11:51:48 2011
In fact lennart and datenwolf just talked past each other. There might be some obsolete or inaccurate information in the datenwolf talk however it does not make the crucial points of his talk irrelevant. The way the desktop evolved in last couple years is really unfortunate.

I will cite his conclusion slide here because I completely agree with it.

Fallacies of contemporary desktop development:
Erection of huge and complex structures
Features given more weight than simplicity and stability
Problems often not properly identified
Problems tackled by throwing even more code at them, instead of fixing proper cause.

Posted by anon at Mon Jan 3 15:15:53 2011
Just as a side note: I'm admin at a university too, managing quite a bunch of Linux Desktop PCs, and while it's true that there are quite a few settings which have to be adjusted for common distributions to be used in this environment, I never had a problem with that. ConsoleKit I find especially useful. It makes it easier to kick/kill idle users. My only criticism with it is, that some of the documented features are in fact not implemented.

Posted by Simon at Mon Jan 3 23:44:56 2011
I just want to say that I really enjoyed your interruptions in the talk! Actually I don't like it when people interrupt others while they are talking. But in this case it was not only alright, but even necessary and important because there were many things said which are simply wrong. Thank you for correcting it and making it clear!

Posted by Andreas at Wed Jan 5 00:20:21 2011
I have just finished watching the video from datenwolf's talk. It was quite embarrassing to see such an uninformed talk at a congress which sets standards - but not this time.
I was good to see/hear your arguments Lennart in contrast to the antiquated ideas of datenwolf. His key criticism of complicating things might be worth discussing, but not in the way he tried to formulate this criticism.
Keep up your good work!

Posted by Wulf C. Krueger at Wed Jan 5 15:20:33 2011
First of all, the talk (I've just watched the recording) was horribly uninformed, indeed. If you look into the subject matter, you can't miss Lennart's name and asking him, thus, would have been quite natural.

Peer review is always a good idea, too, and asking the people whose software you critisise in such a case is smart because in the worst case (them being there!), you can rub their own answers into their faces. ;-)

I don't agree with everything Lennart does or says but he's certainly right on one account: If you don't like it, don't use it or improve it.

When some of us at Exherbo Linux figured out we couldn't achieve what we wanted in another project, we founded our own one and have been happen ever since. We've had our own ideas about an init system, too, and were in the process of implementing it when systemd was announced. I looked into it, worked with Lennart (and working with him is usually easy to very easy (cf. http://lists.freedesktop.org/archives/systemd-devel/2011-January/001003.html)) to get some things that bothered me resolved and the result was that systemd is now our standard init system and we don't need to apply any patches anymore ourselves.

Conclusion: Yes, you can critisise the work of others and you should never be shy about it either but a) inform yourself about the subject matter (and the reasons for certain decisions), b) try to actively work with people to get things resolved (before you give a talk about it) and c) if all else fails, do your own thing.

As for a divide between commercial Linux vendors and "the community" (whatever that may be), there are, of course, different interests around. Commercial entities need to make a profit at some point and, consequently, they have to offer solutions that fit most of their customers. One doesn't have to like that but you can disable the things you don't like or just don't take them up on what they offer.

I've been working with people from RedHat, Novell and others, though, and while we may not always agree, they're usually much more accessible and open to suggestions (not to speak of patches) than any other commercial entity I have to deal with.

That many of us (myself included) have huge egos may not always make things easier but that's just because we're usually right. ;-)

Posted by sHm at Sun Jan 9 16:25:34 2011
Thanks Lennart for your good work to improve linux and bring it to the desktop for the average user. Datenwolf will never understand that 80% are DAU's, 18% are advanced users, sysadmins, ... and 2% are datenwolfs.

datenwolf used the term DAU in his talk
DAU = dumbest assumable user, dümmster anzunehmender User

i prefer the term DAA
DAA = dumbest assumable admin, dümmster anzunehmener Admin