udev (viable) alternatives ?

classic Classic list List threaded Threaded
27 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

tanstaafl-2
On 11/10/2014 10:48 AM, Samuli Suominen <[hidden email]> wrote:
> I wouldn't worry about it at all, there is no way *sys-fs/udev
> ebuild* will ever need systemd. There might be a news item later,
> with instructions on moving to something else, but that's not
> something we are even planning at the moment, so sys-fs/udev is still
> the de facto proper upstream /dev manager.

Well, that sounds reassuring, so thanks very much for this and you're
hard work for all of us non-programmer gentoo users, it is much appreciated!

I guess Lennart was just using words that I read the wrong way... even
now if I re-read his posts, it sounds to me like he is saying 'no more
separate udev without systemd ultimately'... and I know for sure he has
made exactly this comments in the past, but that was admittedly 1 year
or two ago...

Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

Walter Dnes
In reply to this post by Samuli Suominen-4
On Mon, Nov 10, 2014 at 05:48:49PM +0200, Samuli Suominen wrote

> I wouldn't worry about it at all, there is no way *sys-fs/udev ebuild*
> will ever need systemd. There might be a news item later, with
> instructions on moving to something else, but that's not something we
> are even planning at the moment, so sys-fs/udev is still the de facto
> proper upstream /dev manager.

  What worries me is that Lennart has been able to get modifications
done to the kernel, e.g. kdbus.  I know this'll sound paranoid, but how
long before he pushes a patch that requires systemd to run the linux
kernel?

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

Samuli Suominen-4

On 11/11/14 07:20, Walter Dnes wrote:

> On Mon, Nov 10, 2014 at 05:48:49PM +0200, Samuli Suominen wrote
>
>> I wouldn't worry about it at all, there is no way *sys-fs/udev ebuild*
>> will ever need systemd. There might be a news item later, with
>> instructions on moving to something else, but that's not something we
>> are even planning at the moment, so sys-fs/udev is still the de facto
>> proper upstream /dev manager.
>   What worries me is that Lennart has been able to get modifications
> done to the kernel, e.g. kdbus.  I know this'll sound paranoid, but how
> long before he pushes a patch that requires systemd to run the linux
> kernel?
>

I expect systemd-udevd to be migrated into kdbus, which means libudev,
libgudev-1.0 and the
systemd-udevd binary itself will likely need the libsystemd-bus library,
which we will then package
and ship together with sys-fs/udev
Or if systemd-udevd binary starts requiring a running service of some of
the systemd services,
then we will make those available as well and run the from the
udev-init-scripts, or possibly
even adjust sys-apps/openrc to compensate for the inadequaties

Just trying to say, that even with kdbus pending, I'm not worried at all

- Samuli

Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

Canek Peláez Valdés
In reply to this post by Walter Dnes
On Mon, Nov 10, 2014 at 11:20 PM, Walter Dnes <[hidden email]> wrote:

> On Mon, Nov 10, 2014 at 05:48:49PM +0200, Samuli Suominen wrote
>
>> I wouldn't worry about it at all, there is no way *sys-fs/udev ebuild*
>> will ever need systemd. There might be a news item later, with
>> instructions on moving to something else, but that's not something we
>> are even planning at the moment, so sys-fs/udev is still the de facto
>> proper upstream /dev manager.
>
>   What worries me is that Lennart has been able to get modifications
> done to the kernel, e.g. kdbus.  I know this'll sound paranoid, but how
> long before he pushes a patch that requires systemd to run the linux
> kernel?

Then you can take the very last commit to Linus' Git repository before
that hypothetical change, and fork the Linux kernel into whatever
direction your crazy heart desires.

Business as usual in the Free Software world.

Regards.
--
Canek Peláez Valdés
Profesor de asignatura, Facultad de Ciencias
Universidad Nacional Autónoma de México

Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

Rich Freeman
In reply to this post by Walter Dnes
On Tue, Nov 11, 2014 at 12:20 AM, Walter Dnes <[hidden email]> wrote:

> On Mon, Nov 10, 2014 at 05:48:49PM +0200, Samuli Suominen wrote
>
>> I wouldn't worry about it at all, there is no way *sys-fs/udev ebuild*
>> will ever need systemd. There might be a news item later, with
>> instructions on moving to something else, but that's not something we
>> are even planning at the moment, so sys-fs/udev is still the de facto
>> proper upstream /dev manager.
>
>   What worries me is that Lennart has been able to get modifications
> done to the kernel, e.g. kdbus.  I know this'll sound paranoid, but how
> long before he pushes a patch that requires systemd to run the linux
> kernel?

Apologies if this comes across a bit agro, but:

1.  kdbus isn't in the kernel (though that seems likely to happen at some point)
2.  it does sound paranoid
3.  if he pushes a patch that requires systemd I'd be shocked if Linus merged it

Can you find one example of any situation where the linux kernel has
ever required any specific implementation of anything in userspace as
a matter of policy in its 23 year history?  I'm sure you could find
some examples of cases where there just happened to be one de-facto
implementation of something, but even that might be tough with all the
diversity in the linux world.

Linus himself has articulated some of the reasons why kdbus is likely
to get merged.  It fills in a gap in Linux as compared to many
competing operating systems, and it is logical to implement at the
kernel level.  That is generally the criteria for getting stuff into
the kernel, and is basically good software design.  The linux kernel
is all about stable userspace ABIs - if there is only one
implementation of something it is probably because nobody was bothered
enough to write another.

--
Rich

Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

Walter Dnes
On Tue, Nov 11, 2014 at 07:18:30AM -0500, Rich Freeman wrote

> On Tue, Nov 11, 2014 at 12:20 AM, Walter Dnes <[hidden email]> wrote:
> >
> >   What worries me is that Lennart has been able to get modifications
> > done to the kernel, e.g. kdbus.  I know this'll sound paranoid, but how
> > long before he pushes a patch that requires systemd to run the linux
> > kernel?
>
> Apologies if this comes across a bit agro, but:
>
> 1.  kdbus isn't in the kernel (though that seems likely to happen at some point)
> 2.  it does sound paranoid
> 3.  if he pushes a patch that requires systemd I'd be shocked if Linus merged it
>
> Can you find one example of any situation where the linux kernel has
> ever required any specific implementation of anything in userspace as
> a matter of policy in its 23 year history?  I'm sure you could find
> some examples of cases where there just happened to be one de-facto
> implementation of something, but even that might be tough with all the
> diversity in the linux world.

  It might not be an official official requirement, but if the upstream
gets rolled into systemd, then we depend on the "goodwill" of systemd
devs not to go and break anybody else's userspace implementation.
Lennart and "goodwill" do not belong in the same sentence.  How's
systemd-shim working out for Debian???  I'm old enough to remember the
OS/2-versus-Windows wars.  At one point, IBM had Windows 3.1 running
inside of OS/2.  Then Microsoft issued "a minor update" (Windows 3.11)
and it no longer ran inside OS/2.  It took a while for IBM to get
Windows 3.11 running inside OS/2.  That's the kind of hostility that
non-Lennart userspace software faces.
 
> Linus himself has articulated some of the reasons why kdbus is likely
> to get merged.  It fills in a gap in Linux as compared to many
> competing operating systems, and it is logical to implement at the
> kernel level.  That is generally the criteria for getting stuff into
> the kernel, and is basically good software design.  The linux kernel
> is all about stable userspace ABIs - if there is only one
> implementation of something it is probably because nobody was bothered
> enough to write another.

  We Gentoo folks got our wakeup call ( literally
http://lists.freedesktop.org/archives/systemd-devel/2014-May/019657.html )
from the guy who signs himself as "Lennart Poettering, Red Hat"
(nuff said).

  I don't know how long udev will run standalone without systemd.  My
desktop PC is running mdev.  See https://wiki.gentoo.org/wiki/Mdev and
https://wiki.gentoo.org/wiki/Mdev/Automount_USB

  eudev is also an option.  Hopefully, device management doesn't get
forced into systemd.

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

Reply | Threaded
Open this post in threaded view
|

Re: udev (viable) alternatives ?

Tom H-4
On Thu, Nov 13, 2014 at 12:08 PM, Walter Dnes <[hidden email]> wrote:

> On Tue, Nov 11, 2014 at 07:18:30AM -0500, Rich Freeman wrote
>>
>> Can you find one example of any situation where the linux kernel has
>> ever required any specific implementation of anything in userspace as
>> a matter of policy in its 23 year history? I'm sure you could find
>> some examples of cases where there just happened to be one de-facto
>> implementation of something, but even that might be tough with all the
>> diversity in the linux world.
>
> It might not be an official official requirement, but if the upstream
> gets rolled into systemd, then we depend on the "goodwill" of systemd
> devs not to go and break anybody else's userspace implementation.
> Lennart and "goodwill" do not belong in the same sentence. How's
> systemd-shim working out for Debian?

Do you have any reason to think that it isn't working out?

The systemd-shim and cgmanager developers have to play catch up but
I'm using both on Debian and Ubuntu and they're OK.

12