[gentoo-dev] Re: Opinion against /usr merge

[Duncan]

Richard Yao posted on Tue, 17 Jul 2012 17:20:13 -0400 as excerpted:

> The only thing that might require a merge is systemd and it is not in
> @system. If we offered users the ability to choose rc systems, we would
> still be supporting baselayout-1's rc system. If we start now, we should
> bring that back.

Just addressing this little bit.

The first and primary requirement for gentoo support of any rc/init 
system is that there's someone (gentoo dev or not, but baselayout was 
gentoo based so gentoo dev, or at least an advanced gentoo user, would be 
most likely) interested in maintaining it as upstream, and a gentoo dev 
(can be the same person) interested in maintaining the gentoo package/
ebuild.  Gentoo's a community distro build on FLOSS, and if nobody's 
interested in assuming that work/responsibility for either gentoo or as 
upstream, ultimately, it gets tree-cleaned.

Think about it.


Consider things like kde3 and the kde-sunset overlay, too.  That's not 
system init, but you're likely talking a job of similar size to continue 
to support it, with all the necessary initscripts, etc.  kde-sunset is 
what happens when there's sufficiently interested users but no 
sufficiently interested gentoo devs, and/or a break in upstream 
maintainership (tho it eventually continued via the trinity project).


Other (not even officially supported) init systems such as systemd that 
happen to be in our tree have at least that required minimum, and remain 
in the tree.  kde3, despite having an active remaining gentoo userbase 
that continues to maintain it to some degree, didn't have that minimum.

But I can say this.  If there had been a sufficient level of gentoo 
developer interest in maintaining kde3 in-tree, it would have happened.  
Where there's developer interest, the product says around.  There simply 
weren't developers stepping up to maintain it, so it ended up in an 
overlay.


The same thing applies to baselayout-1... and for that matter, 
baselayout-2/openrc.  If there had been sufficient developer interest in 
maintaining a working baselayout-1, it would have happened.  Without it, 
no arbitrary ruling, no "if systemd, then baselayout-1", no "thus saith 
the council", will change it.

Similarly, no wishing, "should be"s, decrees from council, etc, got 
baselayout-2/openrc stabilized, until williamh volunteered to push on it 
(certainly with help from others, but it didn't happen until someone 
stepped up to take personal responsibility for ensuring that it WOULD 
happen) until it got done.


What you're seeing here with the unified / and /usr idea is more or less 
a similar dynamic, tho to quite a lessor extent.  Upstreams are making 
their choices, and the gentoo maintainers of the corresponding packages 
are making /their/ choices.  That's all there is to it.  Upstream's going 
its way, but regardless of that, if there's sufficient interest among 
gentoo devs to guarantee patch development, testing, deployment, further 
testing, stabilization, and bug followup, gentoo WILL continue to make an 
initr*-less separate /usr work.

But one of the things that had (at least until recently, I've seen 
arguments that it's breaking down now, with android, ubuntu, etc, going 
their own way to a greater extent than most before, and android 
especially is big enough to pull it off!) kept Linux from fragmenting the 
way the old Unices did was that while FLOSS ALLOWS forking, it also 
provides significant pressure to ONLY fork where one REALLY feels it's a 
high priority deal.  Otherwise, it's simply less work and less expense, 
to go with upstream.  That's why we have all these distributions, each 
generally different in their own way (and gentoo's certainly rather 
different from the generally binary distros, for sure!), but except for 
the features that a particular distro is emphasizing, that it thinks are 
important enough to be worth the trouble of maintaining that 
differentiation, it tends to stay pretty close to mainline.


Which again, is exactly the dynamic we're seeing here.  What we're going 
thru right now, with all the threads and discussion related to this, is 
simply debating whether, as a distro, gentoo considers this 
differentiation from upstream worth the cost of continued maintenance.  
Which is where the point I made above comes in.  If there's no gentoo 
devs caring enough to make it their job to ensure that support remains, 
the default will simply be to follow upstream, only maintaining the 
minimal patches necessary to continue what gentoo, individually and 
collectively, finds high enough priority to be worth the time and effort 
cost to maintain the differentiation.

And as of now, just as with kde3, the fact of the matter is that despite 
a decent level of interest from users, it doesn't look like we have the 
necessary interest from devs to commit to such support, long-term.  What 
we're seeing is devs committing to maintaining support for the short and 
intermediate (?) future, out perhaps six months to a year or two, but the 
differentiation cost trend beyond that looks to increase quite 
dramatically, and at least at present, we don't have enough developers 
willing to commit to maintaining it beyond that.


And the thing is, no "should"s are going to change that.  If you (and 
other users, personally I'm in the middle, interested but not caring 
enough to really commit to it) want it, there's two ways to get it:

1) Start now, and either convince enough current devs or convince to join 
enough new devs, so when the time comes, that commitment can be made, 
even if it means a significantly increased patch load, to the point of de-
facto or official forking.

2) Get ready for a user-managed overlay, similar to kde-sunset.

Of course the other alternative would be to find a distro more suitable, 
but for a lot of gentooers, that's not an alternative they consider 
viable, as gentoo's close enough to what they want that there really 
/aren't/ a lot of distros even /close/ to suitable, let alone /more/ so.  
Of course, both users and distros change over time, and a lot can happen 
in two years, but...

Then of course there's the "found your own distro" club...  Something 
gentoo's founder, Daniel Robbins, has done more than once, now...  But of 
course just as his current gentoo-based funtoo (and other gentoo-based 
distros, after all, gentoo even claims meta-distro, so gentoo-based distro 
fits right in!), just because you do your own thing doesn't mean you 
can't base on gentoo to whatever degree you want/need, as long as you 
maintain compatible licensing and at least semi-compatible package-
management.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman