Thu Jun 22 12:13:26 CEST 2006
More ressources, more fun
Good news everyone!
I have found a sponsor for a server. Details will follow as soon as that box is operational, it'll provide lots of diskspace on a fast pipe.
That allows all of you who still abuse dev.gentoo.org as SRC_URI and those who want to provide lots of data (testing stages etc.) to host files and services on Yet Another Server.
I will also open this server to powerusers that can show a need, for example Gentoo/HURD and other really experimental Gentoo efforts. If you wish to use this server generously provided by a french hosting provider mail me and I'll see to it.
A big thanks to all that support Gentoo, without you people we wouldn't have one of the best software distributions available.
I have found a sponsor for a server. Details will follow as soon as that box is operational, it'll provide lots of diskspace on a fast pipe.
That allows all of you who still abuse dev.gentoo.org as SRC_URI and those who want to provide lots of data (testing stages etc.) to host files and services on Yet Another Server.
I will also open this server to powerusers that can show a need, for example Gentoo/HURD and other really experimental Gentoo efforts. If you wish to use this server generously provided by a french hosting provider mail me and I'll see to it.
A big thanks to all that support Gentoo, without you people we wouldn't have one of the best software distributions available.
Thu Jun 15 10:33:22 CEST 2006
/dev/urandom
So I've been thinking ... yes, I know, it's dangerous. But still I do it
from time to time.
What are the goals of Gentoo?
These days it looks like everyone has an idea what he/she/it wants from Gentoo, every dev tries to shape it in his image (nothing wrong with that) - but we have no global goal. And because of that we have lots of internal discussion which often boils up into quite personal fights.
Onr good example is the recent "Project Sunrise" discussion. The two camps in this discussion were both trying to do what they thought was best. Their interpretation of best is quite different - Chris (wolf31o2) seems to see a stable, mature distribution as goal while Stefan (genstef) seems to desire a more dynamic and involved community. At a first glance this does not seem to collide, but when you try to motivate contributors by offering a potentially buggy overlay (if it was not buggy it'd already be in our main tree, right?) this of course runs contrary to the whole stable, tested, controlled concept.
How can we fix that?
I see a few options how we can optimize and/or correct the situation in general. All of them are tricky and depend on the cooperation of an overwhelmingly large part of the dev community.
Option 1: Choose a BDFL (Benevolent Dictator For Life). This used to be the case with drobbins, it had some problems (mostly drobbins not being the perfect leader many wanted him to be). With the state of the dev community these days I consider that very unlikely since most of us have difficulties accepting authority ;-)
Option 2: Fork. Split Gentoo into two structurally independent entities that have close social and technical relations. Let's call them "Ricer Distro" and "Enterprise Distro". Ricer Distro can do all the whacky stuff, break and explode as much as it wants. Enterprise Distro picks up any pieces from Ricer Distro it likes and does lots of QA, testing and general things to make its users happy. Hard to do since it duplicates most of the existing infrastructure. Also silly, weren't ~ARCH and overlays meant to provide the possibly b0rked stuff?
Option 3: Keep the anarchy we have now. Have lots of infighting, bleed off developers at a depressing rate. Resist any attempts to change the situation since "it just works"(tm).
Option 4: Get the council and trustees to be more visible. Strenghten devrel and motivate them to interact more with devs when there are problems. Discuss all existing policies and optimize them - we don't need more rules, we need better rules. Thingss like the GLEP process should be evaluated, the idea is good but somehow the implementation sucks or is at least perceived to induce an underpressure. Decide on a "roadmap" so that we all pull in the same direction. Get regular status reports so that we know where we are, otherwise it will be hard to reach any goal. Teach more Quality Assurance, we still seem to have rare cases of non-repoman commits and other Bad Things. Our security team seems to work, getting the QA team up to the same level of organizations is definitely possible and should be amongst our mid-term goals.
Option 5: Everything else I have forgotten to consider.
I tend to prefer a constructive approach that does not alienate more devs than needed, so Option 4 is for now my favourite. If that fails the "fork" option sounds nice, but will drain even more brainpower since there's lots of duplication of effort. The easiest strategy is Option 3, but I assume noone conciously wants that. Options 1 and 5 ... well ... I don't see much potential there, but I'm open to any ideas how to do things.
If I find the motivation to trigger a discussion that will likely be the start of Yet Another Needlessly Long Mailinglist Discussion I might just refine this into an email, but for now I think I need to think some more about it.
What are the goals of Gentoo?
These days it looks like everyone has an idea what he/she/it wants from Gentoo, every dev tries to shape it in his image (nothing wrong with that) - but we have no global goal. And because of that we have lots of internal discussion which often boils up into quite personal fights.
Onr good example is the recent "Project Sunrise" discussion. The two camps in this discussion were both trying to do what they thought was best. Their interpretation of best is quite different - Chris (wolf31o2) seems to see a stable, mature distribution as goal while Stefan (genstef) seems to desire a more dynamic and involved community. At a first glance this does not seem to collide, but when you try to motivate contributors by offering a potentially buggy overlay (if it was not buggy it'd already be in our main tree, right?) this of course runs contrary to the whole stable, tested, controlled concept.
How can we fix that?
I see a few options how we can optimize and/or correct the situation in general. All of them are tricky and depend on the cooperation of an overwhelmingly large part of the dev community.
Option 1: Choose a BDFL (Benevolent Dictator For Life). This used to be the case with drobbins, it had some problems (mostly drobbins not being the perfect leader many wanted him to be). With the state of the dev community these days I consider that very unlikely since most of us have difficulties accepting authority ;-)
Option 2: Fork. Split Gentoo into two structurally independent entities that have close social and technical relations. Let's call them "Ricer Distro" and "Enterprise Distro". Ricer Distro can do all the whacky stuff, break and explode as much as it wants. Enterprise Distro picks up any pieces from Ricer Distro it likes and does lots of QA, testing and general things to make its users happy. Hard to do since it duplicates most of the existing infrastructure. Also silly, weren't ~ARCH and overlays meant to provide the possibly b0rked stuff?
Option 3: Keep the anarchy we have now. Have lots of infighting, bleed off developers at a depressing rate. Resist any attempts to change the situation since "it just works"(tm).
Option 4: Get the council and trustees to be more visible. Strenghten devrel and motivate them to interact more with devs when there are problems. Discuss all existing policies and optimize them - we don't need more rules, we need better rules. Thingss like the GLEP process should be evaluated, the idea is good but somehow the implementation sucks or is at least perceived to induce an underpressure. Decide on a "roadmap" so that we all pull in the same direction. Get regular status reports so that we know where we are, otherwise it will be hard to reach any goal. Teach more Quality Assurance, we still seem to have rare cases of non-repoman commits and other Bad Things. Our security team seems to work, getting the QA team up to the same level of organizations is definitely possible and should be amongst our mid-term goals.
Option 5: Everything else I have forgotten to consider.
I tend to prefer a constructive approach that does not alienate more devs than needed, so Option 4 is for now my favourite. If that fails the "fork" option sounds nice, but will drain even more brainpower since there's lots of duplication of effort. The easiest strategy is Option 3, but I assume noone conciously wants that. Options 1 and 5 ... well ... I don't see much potential there, but I'm open to any ideas how to do things.
If I find the motivation to trigger a discussion that will likely be the start of Yet Another Needlessly Long Mailinglist Discussion I might just refine this into an email, but for now I think I need to think some more about it.
Wed Jun 14 00:23:24 CEST 2006
Mocdor the Bugzinator
I really love this image. It was created by a user as part of the Gentoo Overlay Logo contest. I think it is a nice upgrade to the usual "boring" Gentoo Logo.
And for the last few days Gentoo has felt like burninating anything in its path, the mailinglists have been a flood of ... erm ... people agreeing to disagree in multiple variations of politeness. The split between "old" and "new" devs and "stable" vs. "ricer" is getting more and more difficult to reconcile - I'm not sure how to fix that. If people were less teritorial we could cooperate better, but as christel has already said - within the teams it's quite ok, but as soon as teams interact all hell breaks loose.
The GWN is now almost back to fully working, there are still some organizatorial bugs that need to be ironed out, but I'm quite confident that we'll manage to keep a stable weekly release cycle now.
And that's my random collection of thoughts for today.