Lennart Poettering Takes To Battling Systemd Myths

[BeardedGNUFreak]

Lennart Poettering LOL...

Listening to all the Linux kiddies crying over the clusterfuck that is Linux the development process and architecture thanks to incompetent clowns like Lennart Poettering is absolutely hilarious.

So glad I dumped Linux for a real Unix system like FreeBSD.

[duby229]

This is the dillweed that has been pushing PA down everyones throat. If he pushes out other projects anything like he pushed out pulse then we are all in for some slow ass buggy crap.

I don't think he is smart enough to know the difference between myth and fact.... In his mind if he thinks it then its fact and if you disagree then its myth.... So anything that you disagree with he is going to try and debunk as a myth.

And I don't know why people keep saying that you can build udev by itself.. You can't you have to build the whole thing and then pick out the udev bits.

[Ericg]

Sounds like the preponderance of "everyone else" is "Red Hat".

Arch Linux, OpenSuse, Frugalware, Mageia, Mandrake, Fedora, and with the release RHEL7 you'll have Scientific Linux, CentOS, RHEL, probably Oracle's Unbreakable kernel, and I wouldnt be surprise if SUSE does it as well.

So yes.

Gentoo and Debian have OpenRC & sysV respectively.
Ubuntu and its derivatives use Upstart
Everyone else...systemd.

[ChrisXY]

This is the dillweed that has been pushing PA down everyones throat. If he pushes out other projects anything like he pushed out pulse then we are all in for some slow ass buggy crap.

He didn't push pulseaudio down everyones throat. Ubuntu decided to be an early adopter and packaged a less than ideal config and so it became the quasi-standard way too early.

Also, if someone creates multiple projects which end up being picked up for the default configuration of most bigger distributions because it works better than the alternatives chances are, you're not more intelligent than him.

[Ericg]

Yes, the kernel is bloated. Why do you ask?

Kernel's not really that bloated... I mean could it use SOME trimming, sure, but everything is optional and all drivers are modules.

Unless youre one of those guys that get angry over KMS in which case...really, get over it. KMS works the best, you get the best performance. The only downside to KMS is graphics drivers can cause a kernel crash. But those stability issues should hopefully be sorted out by the open source nature of the drivers. These aren't closed-source blobs that we can't fix problems with. If there's a crash or problem we CAN fix them.

[Ericg]

He didn't push pulseaudio down everyones throat. Ubuntu decided to be an early adopter and packaged a less than ideal config and so it became the quasi-standard way too early.

Also, if someone creates multiple projects which end up being picked up for the default configuration of most bigger distributions because it works better than the alternatives chances are, you're not more intelligent than him.

Agreed. Ubuntu's handling of Pulseaudio was the same handling they did with KDE4. They packaged it early and used non-ideal configurations. Then when it got released in a crappy state everyone complained and that became people's automatic assumptions.

[Rigaldo]

Kernel's not really that bloated... I mean could it use SOME trimming, sure, but everything is optional and all drivers are modules.

Unless youre one of those guys that get angry over KMS in which case...really, get over it. KMS works the best, you get the best performance. The only downside to KMS is graphics drivers can cause a kernel crash. But those stability issues should hopefully be sorted out by the open source nature of the drivers. These aren't closed-source blobs that we can't fix problems with. If there's a crash or problem we CAN fix them.

Well, I can't see how relying in the sheer absence of bugs for stability would be good practice under any circumstances. Preventing a problem is just as important as being able to deal with it once it arrives(cause it usually will).
An approach closer to a micro-kernel regarding the drivers would seem better in my opinion at least. And yes, I've gotten multiple kernel panics/crashes with at least nouveau lately ..
The micro-kernel part I'm talking about is kind of like Windows handles it. It would probably need more changes than just the kernel though, to do the graphics driver part as well though(being able to reload the driver without [basically] rebooting, since it seems Xorg will crash anyway, so all apps will too).

[Alejandro Nova]

<moronic-trolling>

There are many gaps in this list:

1. Systemd IS LENNARTWARE. It's made by Lennart Poettering, so, it's satanic, it will break your bones and your house.
2. Systemd doesn't support audio hardware acceleration. It does audio software mixing... wait... was it PulseAudio?
3. Systemd is PulseAudio. See 1.

</moronic-trolling>

[Rigaldo]

<moronic-trolling>

There are many gaps in this list:

1. Systemd IS LENNARTWARE. It's made by Lennart Poettering, so, it's satanic, it will break your bones and your house.
2. Systemd doesn't support audio hardware acceleration. It does audio software mixing... wait... was it PulseAudio?
3. Systemd is PulseAudio. See 1.

</moronic-trolling>

Actually 2. is true, in that Systemd indeed doesn't support audio hardware acceleration. Damn it Lennart, you've done it again.

[Ericg]

Well, I can't see how relying in the sheer absence of bugs for stability would be good practice under any circumstances. Preventing a problem is just as important as being able to deal with it once it arrives(cause it usually will).
An approach closer to a micro-kernel regarding the drivers would seem better in my opinion at least. And yes, I've gotten multiple kernel panics/crashes with at least nouveau lately ..
The micro-kernel part I'm talking about is kind of like Windows handles it. It would probably need more changes than just the kernel though, to do the graphics driver part as well though(being able to reload the driver without [basically] rebooting, since it seems Xorg will crash anyway, so all apps will too).

I do wish you could reload the driver without rebooting but for now, with how limited performance is already, I can understand why they want to keep in kernel for best performance. Once we get it right kernel-side and we know how to get the most performance out of the drivers, with all the optimizations, then we can start experimenting on things like that. Right now we need as much performance as we can get because we're not getting all that much to begin with (in comparison to amd and nvidia blobs).

As far as nouveau... nouveau just want through a massive re-write kernel-side so bugs/panics will probably happen until all the kinks get worked out with the new code. Make sure to file a bug though if you think you can give enough detail to give the developers an idea of whats happening :P