Fedora defers systemd to F15

Posted Sep 15, 2010 19:39 UTC (Wed) by aleXXX (subscriber, #2742)
In reply to: Fedora defers systemd to F15 by bronson
Parent article: Fedora defers systemd to F15

I agree with ncm. It's a core component, which is intended to replace something which is not broken, so of course there must be a trial cycle.
Everything else would have been unreasonable, no matter what anybody said or thought before.

It's the right decision, and it feels strange that so many people now feel they have to apologize to Lennart, because his code wasn't chosen to replace a core system component in the next release.

I mean, it's an exceptional honor for the author if his code is chosen, it's the default that that's not the case.

Alex

Fedora defers systemd to F15

Posted Sep 15, 2010 20:30 UTC (Wed) by njs (subscriber, #40338) [Link] (3 responses)

There are two totally distinct issues here:

1) The technical question of whether systemd should go in F14. You think it shouldn't. Sure, that's a reasonable conclusion. (It's worth noting, though, that the Fedora people seem to be in a completely different reality than you -- even now no-one there seems to be saying that core components must have trial cycles as a matter of policy, and if you read the transcript than 4 out of 5 FESCO members voted to use systemd in F14, with 1 unsure.)

2) The social question of how contributors are treated in the process of making the technical decision.

I find it very disheartening how many people here seem to think that only the technical decision is important, and it doesn't matter how many contributors are alienated along the way. (Or if some contributors do feel hurt and alienated, then obviously that's their own problem and they should just suck it up, regardless of how reasonable a response it is.)

Fedora defers systemd to F15

Posted Sep 16, 2010 2:43 UTC (Thu) by paulj (subscriber, #341) [Link] (2 responses)

What of the social question of the level of responsibility upon Lennart for accepting that others, notable and experienced people among them, are trying to do their best for what's right for an entire distro and its users (a lot more people than Lennart) as well as him?

His software is *still* going to be shipped in F14, it will still likely end up the default init at some stage.

Given Lennart's tendency to be quite optimistic about the quality and readyness of his code, I am more than happy to have people besides Lennart make last-minute value judgements as to the ship-default state of his code. I don't think a distro should be held hostage to Lennart's feelings.

NB: You can pretty much substitute "any contributing developer" with "Lennart" in the above comment. So there's some general points being made here.

Fedora defers systemd to F15

Posted Sep 16, 2010 4:35 UTC (Thu) by njs (subscriber, #40338) [Link] (1 responses)

Am I being unclear somehow? I feel like I'm repeating myself a lot :-).

No-one's saying that Fedora shouldn't have made the decision it did, no-one's saying that they shouldn't have people besides Lennart making last-minute judgments about shipping his code. (Though if you read that transcript, it may not give you much confidence that they actually made a more informed decision.) And no-one's saying that his feelings should hold anything hostage.

What people are saying -- including the ones who actually made the decision -- is that the process of getting there could have been done in a better way, e.g., by actually deciding and communicating that that was the process they were going to use. They could have gotten to the exact same place without pissing off a very talented (in some ways) contributor, and wouldn't that have been better all around?

I'm sure Lennart will get over it, in the long run it isn't that big a deal. But I'm still astonished at everyone saying "what, nothing happened here, any rational contributor would be happy!"

Fedora defers systemd to F15

Posted Sep 16, 2010 17:09 UTC (Thu) by paulj (subscriber, #341) [Link]

Your post seemed to focus on the social issues of treating a contributor badly by rejecting their code on a last-minute value judgement. Mine asked questions about the responsibility the other way around - for the contributing developer to have some understanding that the integration managers have wider responsibilities and to not throw a tantrum when they don't get their way immediately.

Bill Nottingham, I gather, was very well informed on systemd and in particular on the remaining problems. You'll note that the other voters who reversed from yes to no did so because of Bill's no vote, i.e. recognising his standing (both long standing in RedHat, and on systemd status). Bill had also worked with Lennart about required integration issues.

As others have noted (and I repeated in my comment), there is a need for integrators to be able to apply value judgements. It almost certainly is not possible to codify a tightly defined process for integration. I definitely want my distro's integration decision team to have wiggle room.

More importantly, their ability to do this MUST BE RESPECTED. I'm sorry, but emotional black-mail is not what I want to see be an input to the integration decision making process.