Re: [gentoo-dev] Changing order of default virtual/udev provider

[Daniel Campbell]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 02/09/2016 10:29 AM, Rich Freeman wrote:
> On Tue, Feb 9, 2016 at 12:18 PM, Brian Dolbec <dolsen@g.o>
> wrote:
>>
>> Why must it become yet another shouting match.  And I'm sorry to
>> have to tell you this, but you have been leading the charge in
>> that direction.
>>
>
> Fair enough.  I'll admit that this has been a lot of venting for
> me, and an unnecessary distraction for everybody else.

I don't mind accepting part of the responsibility on that front,
either. I made a few comments that were a bit knee-jerky. Now that
you've clarified your viewpoint I can understand where you come from.
>
> Ulm's post and discussion with xiaomiao on #gentoo-dev have made
> me reconsider my attitude towards this.
>
> The reality is that this change really only impacts non-systemd
> users, so the question really should be what provides the best
> experience for these users.  I think there are legitimate arguments
> for either udev or eudev from that standpoint, but I think we need
> to view them through the lens of what is better for those who would
> prefer not to use systemd (and that isn't limited to openrc, though
> I don't know that this distinction matters).  For those who prefer
> to use systemd it is a moot point.
>
> To ulm's point virtual/dev-manager probably should be a part of
> @system (even if the provider is only the minimal static device
> nodes) - a posix system really isn't usable without anything in
> /dev. Whichever provider is the default for that virtual might be a
> good topic of discussion, but it is not a true dependency for this
> decision, and I should not make it into one.
>
> I still think we'd be well-served to get service managers out of
> @system as well - another good topic of discussion, but again not
> a true dependency for this decision.
>
> Gentoo is and has always been about choice.  That is something I
> will always support.
>
> Personally this has been an area of frustration for me.  I get
> that many have come to Gentoo as a refuge from systemd (as well as
> other things).  I think it is a good thing that we can offer them
> that choice, and I've always supported eudev having a home in
> Gentoo for that reason.
>
> However, I really do think that systemd largely represents the
> future for linux distros.  It will of course evolve over time, and
> perhaps some day it will be replaced, but I think that what it
> changes into is going to look a lot more like systemd than the
> things that came before.  I don't think we really do ourselves a
> service by clinging to the old ways, and I don't think avoiding
> systemd as either a default or a recommended configuration really
> improves our reputation or the user experience.

I see systemd as the future of user-centric distros, and perhaps
dev-centric ones as well, as a way to get the internals out of the
equation and get people to focus on higher levels of the stack.
However, it being the future doesn't necessarily mean it'll be a
*good* future. I see it leading to less and less control over the
lower parts of the software stack, and that will slow the pace of
innovation and possible steps forward. Inertia will set in, and those
who've put themselves in the comfortable position as the leads of
these projects will have considerable influence over the experience of
GNU/Linux users. Despite our differences in preferences, I don't think
either of us wants a situation where a single project owns so much of
the problem space.

We shouldn't be throwing out old ways just because they're old, just
as we shouldn't be adopting the new shiny just because it's modern. I
may refuse to use systemd, but if someone actually needs to make use
of it, their company needs it to do X or Y, it's a highly multi-seat
environment, or whatever else, then use the right tool for the job.
The main issue I've always had with systemd and its followers is their
insistence that it's THE tool, right for ALL jobs, that anything else
is inferior. Such a statement is completely dependent on one's use
cases and software requirements. I'm not saying you've asserted this;
just clarifying a little bit of my own position.

Regarding our reputation, I'm not really sure where that's relevant.
To some degree it's important to retain and attract users, to make our
work more worthwhile and to help expose more bugs so we can fix them.
But who's actually going to complain that, say, Gentoo ships with
OpenRC and eudev, but can be changed to 'pure' systemd at install
time, complete with a profile that sets all the bells and whistles for
them? If anything, we support OpenRC and systemd and pretty much
ignore everything else. To my knowledge we don't even mention runit,
minit, or other options in the handbook. I think the only people who
can rightfully complain about lack of attention or coverage are those
who are using these lesser-known or lesser-used systems. Maybe we can
get some users to step up to the plate and contribute to the
documentation.
>
> And that is the frustration that caused me to lash out a bit on
> this topic.  However, this really isn't appropriate.  This isn't
> holding back systemd, and doesn't really have anything to do with
> systemd at all.  This is about making Gentoo better for people who
> have made the choice not to use systemd, and that isn't something
> any of us should be holding back.  If I want to make things better
> for systemd users there are plenty of areas that I could better
> invest this energy.
>
> So, thanks for bearing with me.
>
Good points, I agree. Thanks for being willing to see the other side,
and dealing with my offhanded remarks. It seems we've reached a pretty
good understanding.

- -- 
Daniel Campbell - Gentoo Developer
OpenPGP Key: 0x1EA055D6 @ hkp://keys.gnupg.net
fpr: AE03 9064 AE00 053C 270C  1DE4 6F7A 9091 1EA0 55D6
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=FYyN
-----END PGP SIGNATURE-----