1 |
On Tue, 6 Aug 2013 19:25:15 +0100 |
2 |
Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> wrote: |
3 |
|
4 |
> On Tue, 6 Aug 2013 14:12:06 -0400 |
5 |
> Alexis Ballier <aballier@g.o> wrote: |
6 |
> > On Tue, 6 Aug 2013 18:41:59 +0100 |
7 |
> > Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> wrote: |
8 |
> > > On Tue, 6 Aug 2013 13:05:07 -0400 |
9 |
> > > Alexis Ballier <aballier@g.o> wrote: |
10 |
> > > > 'occasional unnecessary rebuild' is a big deal since subslots |
11 |
> > > > introduce this regression... |
12 |
> > > |
13 |
> > > Er, no. Subslots just simplify the way accurate slot dependencies |
14 |
> > > are expressed. That's all they are: an alternative to having a |
15 |
> > > larger number of (full) slots plus blockers between some of those |
16 |
> > > slots. |
17 |
> > |
18 |
> > hmmm... no ? |
19 |
> |
20 |
> Er, yes. You may be confusing subslots and :=/:* dependencies, which |
21 |
> are a different feature. |
22 |
|
23 |
Well, ok, but this doesn't relate to what I was writing. Subslot, or |
24 |
slot emulators or whatever, in their current usage with := dependencies, |
25 |
are not fine grained enough for some use cases. Those cause regressions |
26 |
if used improperly. |
27 |
|
28 |
|
29 |
> > > > > There's an easy fix for that: split the package up. |
30 |
> > > > |
31 |
> > > > Great, please start submitting patches at our thousands of |
32 |
> > > > upstreams so that their packages can be split properly. |
33 |
> > > |
34 |
> > > You don't need to patch anything... You just make poppler-stable |
35 |
> > > and poppler-dodgy ebuilds. |
36 |
> > |
37 |
> > And you need to patch it to do it properly. |
38 |
> |
39 |
> You just make the ebuilds install different bits. In effect you |
40 |
> emulate a simple subset of how parts would do it. |
41 |
|
42 |
Which needs patching to be done properly... unless you are suggesting |
43 |
to build it twice and throw away whats not needed just to workaround |
44 |
subslots limitations. |
45 |
|
46 |
|
47 |
> > Or you can do parts/subpackages or subslot dictionaries to express |
48 |
> > that. |
49 |
> |
50 |
> Realistically, parts will never get implemented in Portage. Subslot |
51 |
> dictionaries might be, if anyone ever figures out what they're |
52 |
> supposed to be, but they're a heavy price for package developers to |
53 |
> pay. The question under discussion is whether it's a price worth |
54 |
> paying to avoid an occasional unnecessary rebuild. Since users do far |
55 |
> more unnecessary rebuilds for other reasons anyway, and reducing CPU |
56 |
> usage has never been a goal for Gentoo, I'm not convinced it's worth |
57 |
> caring about. |
58 |
|
59 |
|
60 |
Meanwhile, there's preserve-libs :) |
61 |
|
62 |
Your argumentation is basically 'Other parts are doing it wrong so it's |
63 |
ok to add some more to it'... We're back a dozen emails back, aren't we? |
64 |
|
65 |
|
66 |
> > > > Exercise: Try to update a FreeBSD 6 system (libc.so.6) to a |
67 |
> > > > FreeBSD 7 system (libc.so.7) without some kind of preserve-libs |
68 |
> > > > mechanism. Been there, done that. The flaws of preserve-libs |
69 |
> > > > show up but it maintains a half working system all the way long |
70 |
> > > > that allows you to finish the update. |
71 |
> > > |
72 |
> > > There is no "half working". Something is either correct or it |
73 |
> > > isn't. |
74 |
> > |
75 |
> > Sometimes there is no 'correct'. You are probably using every day |
76 |
> > programs that use heuristics/approximations to solve NP-hard or |
77 |
> > undecidable problems in a 'correct enough' way. |
78 |
> |
79 |
> Dependency resolution is at least NP hard, and we're still solving it |
80 |
> exactly. But you're confusing approximations and feasibility here. |
81 |
|
82 |
It was meant as an example and has nothing to do with dependency |
83 |
resolution. The above exercise is something extreme but that we have to |
84 |
solve; preserve-libs has proven to be correct enough. You have yet to |
85 |
show a correct, in your sense, solution. |