Is udev going to be useless? (see the link)
 

This is an interesting read: https://lkml.org/lkml/2012/10/3/484 .
Opinions?

Forgot this link too https://lkml.org/lkml/2012/10/3/564

I've never liked udev - far too complicated IMO. I've been meaning to spend some time looking at investigating a mdev based replacement, but never gotten around to it.
I hope udev does go away.

This post pretty well summarizes the situation as it is now.

Meanwhile Slackware 14 works reasonably well as it is, and IMHO we don't need one more systemd/udev heated discussion as we have had enough of that recently, see this thread which is hopefully closed

So, just wait and see :cool:

https://lkml.org/lkml/2012/10/2/303 & https://lkml.org/lkml/2012/10/3/484

On a related note whats up with consolekit? The only replacement is systemd's logind.
Are we stuck with it for eternity?

« L'éternité c'est long, surtout vers la fin. » ;)

https://bitbucket.org/braindamaged/udev Forked Udev

talking about useless ...

Perhaps hurricane intteruptions

Devtmpfs, hotplug, and HAL worked far better even if they were a bit dated. This all goes to show that just because something seems better in fundamental aspects doesn't make it better realistically if the developers don't know what the hell they are doing. Someone should tell Linus to add hotplug, HAL, and devtmpfs back into the kernel and ditch udev if the crap doesn't work correctly. Udev is now part of systemd which is a piece of crap as well. The udev fork probably won't fair much better anyway.

Wait... who did Linus call a "Two-Faced Lying Weasel"?

Someone=ReaperX7?

Nope:

http://lkml.org/lkml/2012/10/2/303

Got Linus's email?

hotplug, HAL and udev are not part of the kernel and never were. So, mouthing off about Linus and what he should do will get you nowhere. It always amazes me that the most ignorant people are the first and most eager to make their threatening suggestions...

If you really want to be helpful, first know what you are talking about, then work your way towards contributing some code which will accomplish what you think should be done. But, remember that the open source community has not been just stumbling along, waiting for *you* to come along with your bright ideas.

I think you mean "devfs". devtmpfs is what is used currently in conjunction with udev.

I like devtmpfs, it a natural and logical evolution of devfs. It's the implementation of udev that is letting the side down, and after reading that exchange on lkml it's clearly a point that is not lost on Linus.

Discussion about topics does not need to invite negative comments as such towards myself or anyone else, please reconsider your words when talking to others about things and be polite. Calling anyone ignorant and pointing fingers isn't getting "you" anywhere either, so remember your words and choose them carefully.

I meant to say udev and the original implementations of hal, hotplug, and devfs were supported by the kernel for hardware device management, not exactly part of the kernel.

Ah! Yes, GazL I meant devfs. I forgot devfs had been phased out and replaced with udev's implementation of devtmpfs.

I wonder if Linus has looked into the non-systemd udev fork yet?

It s displayed in posts which are a response to him in the lkml mailing list.

Absolutely agree.
Linus may call Kay a "two-faced lying weasel", but submitted a patch that essentially implements Kay's suggestion. Those high caliber developers know each other for many years and will settle issues without our advices.
Dog barks, caravan is moving.


... who in most cases never written a single line of code themselves but always first to discuss "proper design paradigms", "unix phylosophy" and similar bs.

Interesting development. Hope this gets off the ground:
http://linux.slashdot.org/story/12/1...pers-fork-udev

GazL, you beat me to it--I was going to start another thread, but you dug up this one.

Anyway, I wonder if Pat and the gang (along with the rest of us?) should throw our hat behind the Gentoo effort and actually do real support/coding behind it? The more distros get behind it, the better it should work, and maybe some of the nagging problems with udev can even be fixed!

I hope the fork does go forward and gets adopted by others besides gentoo. However, Kay Sievers seems to have commit privileges which may not be good -since he's the one turned udev over to Lennart Poettering. Also, the new fork seems to follow the upstream changes pretty closely -I wonder what they will do when changes come in which cause new breakages. I'm quite sure that the upstream will, at some point, seriously break compatibility with older versions -which is one of the main problems which has plagued udev all along.

For me, Kay Sievers has joined the shortlist of our greatest villains, behind Ulrich Drepper, Jörg Schilling and Greg Kroah-Hartmann (who created udev and it's traditional lack of both foresight and hindsight). Lennart Poettering is new to the list but is rising fast in the ranks...

they should just bring back devfs

Gentoo developers are also unhappy with udev: http://www.phoronix.com/scan.php?pag...tem&px=MTIzMDU

Sorry, Martinus2u posted the link first :)

Something _is_ going to happen, that we can be certain of, but what? If things continue there will be all sorts of trouble in in big parts of the free software community not related to Linux, but probably also in more specialized cases of Linux far from desktop use.

Somehow, i do not know why yet, i have some ugly feel that this systemd mess can be related to some draconian implementation of secure boot, but that is just some crazy feeling i have.

It seems, that forcing stuff on users is the new FreeDesktop development culture. They forget, that users can leave the GNOME/Pulseaudio/Systemd ecosystem and go for commercial vendors. And without users the remaining hardware vendors will drop support. At the end of the day, the developers sit on their "works perfect for me" implementation, which runs only on a virtual machine. Go figure!

Gentoo Announces Eudev Project -- Its Udev Fork

Yay!!!

udev seems to work reasonably well for me. Aside from that restart I do after boot and logging in to get it back down to one entry in the ps tree. And that, I have less common hardware thing. As I configure this device manually with less of a clue than the developer. Is it me or was MAKEDEV and mknod simpler to understand back in the day? At least it's being looked at and possible solutions are being explored.

Great news! I was a little bit sceptical how eudev turns out, as many forks struggle to stay alive after a while, but it looks like it's doing fine and developers are heading the right direction. Definitely yay!

Udev in Slackware works fine for me now, although it's mainly because of Pat careful choice of udev release. If anything goes wrong with udev in future, we'll see. I would be surprised if anyone revived HAL and other "deprecated" (but still working) tools, nevertheless eudev seems to be good choice for near future.

Interesting - I subscribed to the mailing list just to watch and follow along.

Currently I'm testing a rollback of Slackware 14.0 to udev 165 from 13.37. This has the advantage, that /run can go away too.