KDBUS & Systemd Now Yields A Working System

[phoronix]

Phoronix: Linux KDBUS Now Makes For A Working System

Open-source developers this week achieved a pleasant late Christmas present for Fedora users of having a working system with using the in-development Linux kernel DBus implementation (KDBUS) paired with the latest systemd code can now yield a booting system...

http://www.phoronix.com/vr.php?view=MTU1NDM

[mark45]

Smells like KDBUS no later than Fedora 22.

Yes, but non-devs won't notice a difference anyway.

[Ericg]

Yes, but non-devs won't notice a difference anyway.

Evolutionary step, not a revolutionary one. Not a bad thing.

[ghexsel]

Evolutionary step, not a revolutionary one. Not a bad thing.

Actually, this is apparently motivated by performance, so I suppose end-users may notice it (or measure it).

[Teho]

Actually, this is apparently motivated by performance, so I suppose end-users may notice it (or measure it).

To my understanding it's more about reliablity and security than performance. It's also one of prerequisite for the sandboxed applications for Gnome and Linux in general. There's some info on LWN.net too.

[mether]

Actually, this is apparently motivated by performance, so I suppose end-users may notice it (or measure it).

Performance is one of the factors but for the large majority of use cases, performane of D-Bus does not matter as much compared to the other features that it does provide. Kdbus is instead motivated by other factors outlined at http://lwn.net/Articles/551969/, one of which is security and integration with kernel level features better.

[Annabel]

Oh god, please no
DBus was already bad but this is worse

[89c51]

Oh god, please no
DBus was already bad but this is worse

After that i expect a commend about Lennart touching (or even looking at) the code.

[JS987]

To my understanding it's more about reliablity and security than performance. It's also one of prerequisite for the sandboxed applications for Gnome and Linux in general. There's some info on LWN.net too.

Moving code from user space into kernel will make security worse.

[zester]

First off D-Bus sucks, It really sucks. Its over complicated and boated. Now we have it implemented in the Kernel
with a weird api. IPC is useful and fun but its rarely utilized to any real capacity on Linux because D-Bus sucks.

Then there is systemd, its massively bloated, usage is fairly simple but its architecture is extremely complex, eventually this will
suffer from the Fogbank effect.

Lastly we have pulseaudio - This is just bad for audio in general. Those in Pro Audio and Gamers normally disable it
and use Jack or OpenAL/Fmod directly.


If you advocate for any of those three your an idiot, and do everyone a favor stay out of the kitchen because we already have to many shitty cooks.!!!!