1 |
On Mon, 5 Aug 2013 12:22:32 -0400 |
2 |
Alexis Ballier <aballier@g.o> wrote: |
3 |
> On Mon, 5 Aug 2013 17:13:49 +0100 |
4 |
> Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> wrote: |
5 |
> > On Tue, 06 Aug 2013 02:03:28 +1000 |
6 |
> > Michael Palimaka <kensington@g.o> wrote: |
7 |
> > > > How often does this situation even come up? If 9/10 times the |
8 |
> > > > libraries are set up as maintainers expect them to be, it is |
9 |
> > > > probably better to deal with the odd unnecessary rebuild until |
10 |
> > > > somebody spots it, rather than going without support for slot |
11 |
> > > > operator deps. |
12 |
> > > |
13 |
> > > With respect, "good enough" is not a very high standard to aim |
14 |
> > > for. In my opinion, adding unnecessary subslot dependencies is no |
15 |
> > > different to adding overly-wide dependencies. |
16 |
> > |
17 |
> > There's a world of difference between a horrible breakage and an |
18 |
> > occasional unnecessary compile. If users are concerned about how |
19 |
> > they spend their CPU time, they're using the wrong distribution. |
20 |
> |
21 |
> there is something wrong in the way its done if there are |
22 |
> 'occasional unnecessary compiles' |
23 |
|
24 |
Not really. There's a tradeoff between dependencies that are |
25 |
occasionally too strict, and dependencies that are horribly |
26 |
complicated (see "subslot dictionaries"). |
27 |
|
28 |
Gentoo already favours lots of unnecessary recompiling over additional |
29 |
complexity. For example, for many revbumps, it would be possible to |
30 |
only rebuild a small part of the package and replace a few files rather |
31 |
than the whole thing. But writing ebuilds capable of doing so would |
32 |
involve more developer work, and would be more prone to screwups, so |
33 |
such a feature isn't offered. |
34 |
|
35 |
So all this fuss over "unnecessary compiles" is misplaced. Gentoo |
36 |
simply isn't the right distribution to use if minimising compile time |
37 |
is a goal. |
38 |
|
39 |
> 'horrible breakage' is mitigated by preserve-libs and running |
40 |
> @preserved-rebuild as soon as possible has the same end result |
41 |
> avoiding useless rebuilds. |
42 |
|
43 |
But preserve-libs introduces breakages and security holes. The point of |
44 |
slot operator dependencies is to replace that with a reliable feature |
45 |
that doesn't rely upon guesswork and voodoo. |
46 |
|
47 |
-- |
48 |
Ciaran McCreesh |