Gentoo Archives: gentoo-dev

From: Jeremy Olexa <darkside@g.o>
To: gentoo-dev@l.g.o
Subject: Re: [gentoo-dev] New ebuild metadata to mark how robust the package is?
Date: Fri, 16 Oct 2009 23:48:51
Message-Id: 4AD90660.5040903@gentoo.org
In Reply to: [gentoo-dev] New ebuild metadata to mark how robust the package is? by Daniel Bradshaw
1 Daniel Bradshaw wrote:
2 > Hi all,
3 >
4 > It occurs to me that my work flow when doing updates follows a fairly
5 > predictable (and probably common) pattern.
6 > The obvious next step is to wonder why no one though of automating it...
7 >
8 > When doing updates I tend to look through the package list and classify
9 > things based on how likely they are to break.
10 > Some packages, like findutils, are pretty robust and generally just get on
11 > with working.
12 > Other packages, like apache and ssh, need are more fragile and need plenty
13 > of configuration.
14 >
15 > Packages from the second group want emerging on their own, or in small
16 > groups, the better to keep an eye out for notices about things that might
17 > break, to update configs, and to check that they're running happily.
18 >
19 > Once the update list is reduced to packages from the first group it's
20 > fairly safe to run emerge -u world and not worry about things exploding
21 > too badly.
22 >
23 >
24 > So as I say, it occurs to me that most people probably follow some
25 > variation of this selective upgrade method.
26 > It might be handy to have some kind of metadata in the ebuilds that can be
27 > used to indicate a package that is "demanding".
28 > Then that flag could be used to highlight the package on a dep tree, or
29 > optionally to block the emerge unless the package is specified explicitly.
30 >
31 > `emerge -vaut @safe` would be kinda useful.
32 >
33 > Just a thought.
34 >
35 > Regards,
36 > Daniel
37 >
38
39 I am seeing a trend here because this email aligns with the thoughts on
40 the openrc email thread a few days ago. That being said, no clue how to
41 implement it. Actually I think that marking a package as "demanding"
42 would be the more useful than "safe" because probably 95+% of packages
43 are "safe"
44
45 But, as with a request for an indicator for compile times[1], I think
46 this proposal would be a failure in general because of subjective
47 opinions between people. I would consider apache/lighttpd as being
48 "safe" after initial configuration as long as you don't automatically
49 etc-update. But you or someone else would not think it was safe.
50 Therefore, nice in theory but probably wouldn't work in practice.
51
52 Nice idea, keep them rolling and I'm not trying to kill the thread. :)
53
54 -Jeremy
55 [1]: http://bugs.gentoo.org/288193