[gentoo-dev] Re: AMD64 Arch Testers needed urgently

[Duncan]

M. J. Everitt posted on Fri, 15 Dec 2017 01:25:38 +0000 as excerpted:

> 1) Gentoo isn't really interested in having a 'stable' tree or it would
> already be happening. As such, why not cut the Gordian knot, declare
> that this is not something that will happen [soon] and let users make
> their own choices. The [majority of] developers already seem to have ...

The (general) argument for keeping stable seems to consist of two points 
which together have always (to the present at least) tended to 
overwhelmingly tip things in favor of keeping it.  Indeed, either one 
alone can do so, even when the other one is discounted for some reason 
(like some people not putting value in that point).

a) The world of FLOSS isn't a zero-sum game.  People tend to contribute 
where they have interest (whether it's theirs personally or that of their 
employer), and attempting to ban whatever "wasted effort" project in 
favor of what they "should" be contributing to seldom results in /that/ 
much more effort going to the "favored" project after all (and may result 
in less), because if people were interested in that they'd already be 
contributing to it, and the fact that they aren't, or aren't very much, 
tends to mean their interest is elsewhere, and they'll either go 
somewhere else or simply stop contributing as much if they can't 
contribute to what they were interested in, in the first place.

By this point (again, keep in mind we're talking /general/ here, the 
apparent current exception is addressed below), stable exists because 
it's of enough interest to certain contributors that it /continues/ to 
exist, and however much people such as myself (for I'm simply not 
interested in stal^Hble) might consider it a waste of time, killing it 
would be very unlikely to result in invigorating the ~arch I'm primarily 
interested in.  In fact, likely the opposite due to less people as a 
whole contributing.

That's the idealistic point, now perhaps the more practical one...

b) As a point of fact many gentoo sponsors, including those providing 
hosting and/or paying a few gentoo devs to actually do the gentoo work  
they'd very possibly be doing in their spare time anyway (even if it's 
just payment for the 20% aka one day a week open source community 
contribution that's a thing in the tech world), are primarily interested 
in gentoo-stable.

One big and public example is Google, which uses a gentoo base for its 
ChromeOS product.  I'm not privy to details, but from what I've read it 
seems they start with snapshots of gentoo stable, then stabilize them 
further, often feeding their additional patches back upstream to gentoo 
(of course as gentoo feeds stuff back upstream as well, but for some 
things, gentoo /is/ the upstream).  Tho of course if they have reason to 
they can and do pick individual ~arch packages to supplement the stable 
base as well, but the point is, they're interested in /stable/.

And google sponsors, directly or indirectly (some gentooers apparently 
work at unrelated google jobs), a number of gentoo devs, in addition to 
the testing and patches they feed back to us.

Similarly, some of those in our mirror network, for instance, are 
commercial hosters doing it because some of their customers wanted gentoo, 
and providing that local mirror for them, but usable by the public as 
well, is a nice point of convenience to keep those customers /as/ 
customers.

And that's commercial server biz, where the general rule is if it's not 
broke, don't try to "fix" it and in the process risk your view/income/etc 
stream.  So a slow stable actually works well there.


So while I'm personally very much an ~arch and even live-git-build on 
some sets of packages (kde-frameworks/plasma/apps, primarily, with 
occasional others) type of guy and don't /personally/ see much practical 
use for stal^Hble, I certainly see how keeping it makes gentoo bigger and 
better for all of us, including those of us that don't use it personally.


> 2) Whilst there has yet another fine bike-shed emerged on the subject, I
> have only seen one volunteer willing or capable to actually take on
> implementation of anything that has been discussed on this thread. As
> such, you can talk all you like .. nothing will happen until somebody
> actually *does* something ..

Given the context above, What seems to be the problem here is that the 
people that /had/ been interested in, and thus contributing to, gentoo/
amd64-stable, basically the (a) point folks above, seem to have moved on 
to other things, whether inside gentoo or out.

But the (b) point surely remains, so we have a problem.

One could argue that in that case the sponsors should sponsor amd64-
stable folks, but it's not generally that direct, and even if it were, 
getting that worked into the sponsorship pipeline would take time.


I don't have a solution (my reason for posting was to point out that it's 
not as simple as just dropping stable, not really to provide an answer I 
don't have), but can note that I believe there's two people now 
volunteered for it, and of course people have to be aware of it before 
they can realize their personal stake and thus interest in it, thus this 
thread... Even if not enough by itself, you gotta start somewhere, and 
there's at least the two interested, now...

-- 
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