Gentoo Archives: gentoo-dev

From: Patrick Lauer <patrick@g.o>
To: gentoo-dev@l.g.o
Subject: Re: [gentoo-dev] Re: Training points for users interested in helping out with ebuild development
Date: Tue, 12 May 2009 06:35:46
Message-Id: 200905120835.41799.patrick@gentoo.org
In Reply to: Re: [gentoo-dev] Re: Training points for users interested in helping out with ebuild development by Ciaran McCreesh
1 On Tuesday 12 May 2009 00:31:36 Ciaran McCreesh wrote:
2 > On Mon, 11 May 2009 23:17:32 +0100
3 >
4 > George Prowse <george.prowse@×××××.com> wrote:
5 > > An equilibrium seems to have been reached which currently works.
6 >
7 > An equilibrium has been reached, agreed, but that it works is up for
8 > debate. There is a strong argument to be made that preserving the
9 > equilibrium will keep Gentoo the way it is now -- delivering at best the
10 > same user experience now that it was several years ago, in an
11 > increasingly difficult and more competitive environment.
12
13 At times I do wonder ...
14
15 [long-winded rant following]
16
17 I've been using a few different distros over the last years. Mostly because
18 people claimed that Gentoo is too hard (hey, it has a Big Friendly Manual and
19 Ubuntu has a GUI installer! Ubuntu must be easier!) or didn't want to change
20 their ways ("I've always used Debian Lenny. Why should I change now!")
21
22 What I realized after some time is that we're ahead of the curve. Baselayout 1
23 is neat, but OpenRC is awesome. Machines rebooting in the time it takes some
24 other init systems to not properly restart a service and such funny things.
25 Service dependencies. A proper network config that doesn't make you bite the
26 edge of the table in frustration.
27
28 Then there's package management. (Your favourite topic, I guess, because you
29 want to keep complexifying it until one needs a PhD to write an ebuild [which,
30 in a way, would be quite ironic])
31 Ever tried getting a newer version of PHP with _that_ feature enabled that the
32 distro maintainer didn't like and thus disabled? Whee. Fun.
33 And on larger installations you usually need slightly customized packages, so
34 these jokers build things from source. Manually.
35 Makes it easy not to get updates too ...
36
37 Then you get the bonus features - you can have multiple versions of KDE
38 installed at the same time. There's quite a lot of packages (some not that
39 well maintained, but that's hard to avoid) and lots of unofficial and semi-
40 official overlays that have most things you need.
41
42 So I'd say we're in a rather good position.
43
44 And now you say "delivering the same user experience" ...
45 ... ignoring the tons of new features and things that have happened. You're
46 being dishonest again in an attempt to make us look like baboons. Two thirds
47 of the new features grew on your compost heap (and half of these features we
48 didn't even want, but after about three years of you pushing them at every
49 opportunity people are getting so demotivated that they are willing to let you
50 have one feature if you just STOP WHINING for more than 10 minutes)[GLEP55,
51 for example - there's about 8 people that want it, but those keep bringing it
52 up at EVERY opportunity. It's still a fundamentally stupid idea that doesn't
53 solve any problems, and the claim that it might solve problems we have in the
54 future is quite asinine because we can do the changes then, _if_ the
55 theoretical problems actually become an issue, without messing up most
56 everything now for some hypothetical gain that has not even conclusively shown
57 ...]
58
59 People have forked Gentoo with the goal of "making things better", and look
60 where it leads them - most of them turn into a passive downstream of gentoo,
61 absorbing bugfixes with a day to a month delay until bugfixes make it in. And
62 those that don't stay passive slowly collapse until they are nothing more than
63 a shiny webpage.
64
65 People refuse to learn from that, but the "lessons" are quite obvious:
66
67 - We're not in a bad shape, dying or dead. We don't intend to.
68
69 - Progress doesn't have to be disruptive features. Use-deps are a behind-the-
70 scenes improvement that few users hit directly, so most aren't even aware of
71 the improvements happening
72
73 - More complex doesn't mean better.
74 "Perfection isn't when you cannot add more things but when there are none left
75 to remove" or how that quote went. You know what I mean. Rewriting the init
76 scripts in XML might be what some call progress (now you can verify 'em!), but
77 it doesn't actually add any value and complexifies things in a bad way
78
79 - Repeating a lie can make it true, if you repeat it long enough. Worst case
80 you just have to wait until everyone who disagrees dies of old age.
81
82 So anyways, just felt the need to rant a bit too. Can't let you keep a
83 monopoly on that, eh?

Replies