Systemd 214 Comes "Stuffed With Great New Features"

[bkor]

Ya' know what this means? Less code in PID1!

Plus those generators could possibly run in a more restricted environment.

[bkor]

Version 344: We know you like classic games, that's why we have included Tetris directly in systemd. Why download a separate piece of software when you can have it all in systemd?

Version 478: Are you sick of buying AAA games and expending money on them? Lucky you! Now you can play Metro 3150 and other AAA games directly from systemd. Are you ready? There we go.... Don't forget to use your verichip reader when logging in!

Version 666: Now you don't need to install anything else, everything is included in systemd 666 black hole edition. You knew resistence is futile right from the start.

systemd is meant as a basic building block for Linux. This is explained many times. At every presentation that they hold, in discussions, articles/blogposts/etc. A basic building block doesn't mean games, that's entirely silly. I do see it to have pretty much anything needed to bring the system up (basic infrastructure). Systems are now dynamic, so that also includes handling of any events (so when you plug in an usb stick).

If you read the announcement they're also changing how things are handled. When it'll reread partition tables, allowing tools to lock a disk to avoid any bad interactions, then changing fsck to do the right thing, etc. Changes such as this one will eventually lead to a reliable system, because all the different parts have been checked and ensured that they work together. And not just work, but really aligned.

At the same time this work is happening you're making silly arguments about games being included...

[Vim_User]

It is quite interesting that OpenBSD have started cloning systemd features (they have a GSoC project), and there can be no doubt that BSD will get a modern init system down the road too, heavily inspired by systemd.

Nonsense. They are just cloning the things needed by Gnome to run. They won't touch their init system at all.

[CTown]

If it's in systemd, how could it be gnome-only?

being able to rebuild /var if it's empty at boot time is my favourite! Perfect for live CDs or fat clients. To date, I had to use a self made script which copied specific stuff over. But this was slow...

That's a good point. Thanks.

[Luke]

If non-init features are worse don't use them

Not that it's exceptionally good at replacing init but it's much worse at everything else.

Where systemd is doing something other than replacing init (as from Upstart or sysVinit), you can ignore or replace that module if you have a better way. Example: the current systemd cryptsetup module (in Ubuntu) is buggy in that is will often call the passphrase more than once for the same volume, actually unlocking it on the first call. I know this because if you enter a blank passphrase on subsequent calls it works fine. I suspect a race condition between looking for the unlocked volume and actually unlocking it. Remember that systemd seeks to do everything at once for maximum speed, this is the very definition of parallelizing the boot process.

Well, I already had my own cryptsetup code anyway for boot time, so I figured the changes to my program would be minimal. They were: it installs in a different place, and some passphrase-handling and messaging code had to change, and it is started by systemd. That's about it, and when plymouth is running it calls plymouth ask-for-password directly rather than having systemd call it to shorten the code path and reduce the potential attack surface. When Ubuntu decided to switch, I decided to port my code over instead of pin the whole init system and whine about it. To do this I had to switch to systemd ahead of Ubuntu, so as to have something to test it on. It works, and systemd works fine in my systems, so it stays.

To boot systemd back off of my systems will require someone to come up with a better (faster and more robust) init system that becomes compatable with Debian or Ubuntu based systems, or hard evidence (with analysis of source code) of a back door dropped into systemd but not into upstart or sysVinit. Anyway, I've enjoyed playing with systemd for the past couple months, it's not some unhackable monolith full of stars.

[interested]

Nonsense. They are just cloning the things needed by Gnome to run. They won't touch their init system at all.

Not yet, but as you know, systemd is more than an init system. The best proof for that this GSoC project is in fact a clone of a part of systemd, is its statement, where the poor developer spend all the time explaining how this project isn't systemd related even though it is exactly like systemd. All that denial wouldn't be needed if this GSoC project couldn't be used as building block for a systemd clone.

It will probably take the BSD guys a few years to have cloned the parts of systemd they want. Sure, they will pull their sad faces and claim that this new init system was forced upon them from external circumstances, but also claim it is entirely unlike systemd, because it is called "SystemB" (as in SystemBSD) with capital letters. And that the "SystemB" lead developer, "Theo de Raadt" from OpenBSD, unlike Lennart Poettering is known for being a humble, ego-less guy with a mellow, easy-going attitude towards everybody.

[DrYak]

And that the "SystemB" lead developer, "Theo de Raadt" from OpenBSD, unlike Lennart Poettering is known for being a humble, ego-less guy with a mellow, easy-going attitude towards everybody.

At least, you're looking forward for a website called "Systemd Valhalla Rampage", where they fork Systemd into SystemB and rip out any security suspicious bug they find in Poettering's code.

[prodigy_]

It will probably take the BSD guys a few years to have cloned the parts of systemd they want. Sure, they will pull their sad faces and claim that this new init system was forced upon them from external circumstances, but also claim it is entirely unlike systemd, because it is called "SystemB" (as in SystemBSD) with capital letters. And that the "SystemB" lead developer, "Theo de Raadt" from OpenBSD, unlike Lennart Poettering is known for being a humble, ego-less guy with a mellow, easy-going attitude towards everybody.

Oh, it's really funny that how UNIX in general and BSD in particular predate Linux by decades but systemd kids still post bullshit implying that BSD takes ideas and code from Linux and not the other way around. Well, at least let's remember it's all free software and thus let's not degrade to who-stole-what nonsense, OK?

And as for Theo de Raadt, he earned our respect and the right to have his attitude through years and years of developing actually useful things (OpenSSH, anyone?) and producing high quality code with top-notch security and reliability. That's something Lennart Poettering could only dream about.

[pixo]

Oh, it's really funny that how UNIX in general and BSD in particular predate Linux by decades but systemd kids still post bullshit implying that BSD takes ideas and code from Linux and not the other way around. Well, at least let's remember it's all free software and thus let's not degrade to who-stole-what nonsense, OK?

And as for Theo de Raadt, he earned our respect and the right to have his attitude through years and years of developing actually useful things (OpenSSH, anyone?) and producing high quality code with top-notch security and reliability. That's something Lennart Poettering could only dream about.

I have a trollish mood so lets do the pointing -> Soalris 10 SMF (January 31, 2005)
The idea behind systemd is old and proven.
I love how i can check which services started and which died without digging through messages and other logs trying to find why something is not up.
And being able to create dependencies is also great.

[interested]

Oh, it's really funny that how UNIX in general and BSD in particular predate Linux by decades but systemd kids still post bullshit implying that BSD takes ideas and code from Linux and not the other way around. Well, at least let's remember it's all free software and thus let's not degrade to who-stole-what nonsense, OK?

And as for Theo de Raadt, he earned our respect and the right to have his attitude through years and years of developing actually useful things (OpenSSH, anyone?) and producing high quality code with top-notch security and reliability. That's something Lennart Poettering could only dream about.

Theo or OpenBSD didn't invent SSH. OpenSSH was a clone of an already existing implementation made by SSH inventor Tatu Yl?nen in 1995. This just underscore my point, that OpenBSD will also clone systemd down the road. It is only a matter of time.

When OpenBSD finally gets an init system that separate config statements into structured text files, from executable code, I will try to restrain myself from trolling every "SystemBSD" thread with statements on how it is the UNIX way to have executable config files, where code and declarative config statements are mixed together in a free form text file:-)

Personally I have no problem with Theos, lets say, colourful personality., but I must say I find it funny when BSD users jumps Linux threads, saying systemd is bad because they have issues with Lennarts personality.

I also find it peculiar when they say Poettering is a bad programmer, when their lack of knowledge about even simple systemd functions show they have never glanced at its source code. Poettering is in fact a very bright star programmer, who despite his young age, already have made his mark with his software being used on millions of devices. Stuff like Avahi, PulseAudio and systemd just works, and they work well and solve real world problems. I think the slandering of his abilities are just an example of the usual failed strategy of trying to "poison the systemd well". It will get you nowhere and just reflect badly back at BSD.