KDBUS Is Taking A Lot Of Heat, Might Be Delayed From Mainline Linux Kernel

[phoronix]

Phoronix: KDBUS Is Taking A Lot Of Heat, Might Be Delayed From Mainline Linux Kernel

Earlier this week I wrote about how it looked like KDBUS would be included in the Linux 4.1 kernel given the pull request sent to Linus Torvalds by Greg Kroah-Hartman. However, since that pull request, KDBUS is taking a lot of heat and there's calls for it to be postponed from mainlining...

http://www.phoronix.com/scan.php?page=news_item&px=KDBUS-Taking-Heat

[mark45]

From the article:

With the kernel not being allowed to break its interfaces with user-space, there's much interest in making sure the KDBUS design is right before its cemented within the kernel as a core feature.

Is that related to the shitty Linux idea of not breaking compatibility with anything ever? Sticking to any extreme is pretty much always a shitty idea.

[dh04000]

From the article:

Is that related to the shitty Linux idea of not breaking compatibility with anything ever? Sticking to any extreme is pretty much always a shitty idea.

I am NOT a kernel expert, but even I know that you are completely wrong.

[mark45]

I am NOT a kernel expert, but even I know that you are completely wrong.

Dude, either tell on what and why or shut up, and I am not a kernel expert NOR AN ASTRONAUT!

[pgoetz]

From the article:

Is that related to the shitty Linux idea of not breaking compatibility with anything ever? Sticking to any extreme is pretty much always a shitty idea.

No, it's not that. People have concerns about the design of the protocol (i.e. they think it's poorly thought out). That the kernel won't break ABI is the excuse for not taking it in without a more thorough vetting. In the end, this will result in a better subsystem for linux.

[Guest]

I am NOT a kernel expert, but even I know that you are completely wrong.

You know right, probably a troll attept.

[mark45]

No, it's not that. People have concerns about the design of the protocol (i.e. they think it's poorly thought out). That the kernel won't break ABI is the excuse for not taking it in without a more thorough vetting. In the end, this will result in a better subsystem for linux.

Now this is a reasonable answer.

[jo-erlend]

From the article:

Is that related to the shitty Linux idea of not breaking compatibility with anything ever? Sticking to any extreme is pretty much always a shitty idea.

There are differences in compatibility. One thing is to drop support for 80386. Another is to break support with user-land. Imagine waking up one morning and your Gnome is broken because you needed to upgrade your kernel in order to get some new drivers.

[YoungManKlaus]

No, it's not that. People have concerns about the design of the protocol (i.e. they think it's poorly thought out). That the kernel won't break ABI is the excuse for not taking it in without a more thorough vetting. In the end, this will result in a better subsystem for linux.

but ... shouldn't be the protocol be the same as "regular" dbus? I.e. act as a drop-in-replacement? Which in turn means the protocol is not up for debate ...

[blackout23]

but ... shouldn't be the protocol be the same as "regular" dbus? I.e. act as a drop-in-replacement? Which in turn means the protocol is not up for debate ...

It is. I saw someone propose that kdbus should break dbus so that applications will have to adopt the new potentially better design.