Gentoo Archives: gentoo-dev

From: Rich Freeman <rich0@g.o>
To: gentoo-dev <gentoo-dev@l.g.o>
Subject: Re: [gentoo-dev] rfc: adding sys-apps/iproute2 to the @system set
Date: Sat, 06 Sep 2014 13:37:51
Message-Id: CAGfcS_=D9cYUVA5usvQP_Kop7JYMo2FgoTUh1+0+PEgek4gdcQ@mail.gmail.com
In Reply to: Re: [gentoo-dev] rfc: adding sys-apps/iproute2 to the @system set by "Anthony G. Basile"
1 On Sat, Sep 6, 2014 at 9:23 AM, Anthony G. Basile <blueness@g.o> wrote:
2 >
3 > I'm not sure we can get rid of python2 and have only python3, but if that's
4 > possible, absolutely punt it! The bloat I'm talking about includes size,
5 > but more importantly, I'm concerned about cpu time. When building on a minor
6 > arch where your CPU speed is 600 MHz and you only have 256MB of ram (and
7 > lots of slow swap to help for monsters like gcc-4.8), you feel the bloat in
8 > days of waiting.
9
10 The other issue is the parallel build issue. @system and its deps
11 can't be built in parallel. That means a penalty for every update to
12 any of those packages for life.
13
14 Any kind of actual end-user application that goes into @system greatly
15 compounds this problem. Applications tend to have lots of
16 dependencies.
17
18 There isn't much question that stuff like rsync and nano (via the
19 editor virtual) should be in the stage3 just so that we're not ripping
20 our hair out during installation. However, they really don't need to
21 be part of the system set. How many packages really need to depend on
22 an editor (and I'm talking linking and other technical issues that
23 affect builds - not practical use)? Of course, people probably don't
24 want to unmerge the last text editor or rsync from their system which
25 is why it doesn't hurt to have some kind of mix-in that defines
26 minimally-useful stuff like this all the same, but which separates it
27 from the practice of not declaring dependencies.
28
29 I'm sure all of us have our favorite utilities that we put on every
30 Gentoo install we do (tmux/screen, atop, vim, etc). The problem is
31 that once you go down that road we end up in endless debates. If we
32 instead ask questions like "what are all the packages which >30% of
33 the tree would otherwise have to depend on if not in @system?" or
34 "what is the minimum set of packages that need to be preinstalled to
35 build anything else in the tree?" we have unambiguous questions that
36 have unambiguous answers.
37
38 --
39 Rich