1 |
On 16/08/2013 02:00, »Q« wrote: |
2 |
> On Thu, 15 Aug 2013 10:16:25 +0200 |
3 |
> Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
4 |
> |
5 |
>> On 15/08/2013 09:30, Nikos Chantziaras wrote: |
6 |
>>> When running: |
7 |
>>> |
8 |
>>> emerge @preserved-rebuild |
9 |
>>> |
10 |
>>> and nothing is found that needs a rebuild, but: |
11 |
>>> |
12 |
>>> revdep-rebuild -i |
13 |
>>> |
14 |
>>> does find something, should it be considered a bug with the |
15 |
>>> preserved-rebuild mechanism and be reported, or is this expected to |
16 |
>>> happen from to time? |
17 |
>> |
18 |
>> The latter, it happens from time to time. I see it here about once |
19 |
>> every 2 months or so (i.e. seldom). I suppose this is the best we can |
20 |
>> expect, seeing as how it all works: |
21 |
>> |
22 |
>> @preserved-rebuild tries to remember everything that uses everything |
23 |
>> and detect changes, this will never be 100%, |
24 |
>> revdep-rebuild actively looks for brokenness and still sometimes gets |
25 |
>> it wrong (dynamic plugin modules anyone?) |
26 |
>> |
27 |
>> The specifics depend on the exact package. Maybe also file a bug so |
28 |
>> Zac can look it over just in case there in a useful tweak he can make |
29 |
> |
30 |
> I thought the use of subslots was supposed to make revdep-rebuild |
31 |
> obsolete someday. |
32 |
> |
33 |
> |
34 |
|
35 |
I doubt that day will ever arrive. |
36 |
|
37 |
subslots are a way for devs to tell portage that the condition |
38 |
revdep-rebuild detects can happen, and to rebuild stuff as part of the |
39 |
emerge. |
40 |
|
41 |
So what happens if the subslot is wrong or missing in the ebuild? |
42 |
Portage can't detect that and humans leave such stuff out all the time. |
43 |
|
44 |
So I think revdep-rebuild is going to stick around for ever, even if it |
45 |
is just in a double-check-stuff and I-got-your-back type role |
46 |
|
47 |
-- |
48 |
Alan McKinnon |
49 |
alan.mckinnon@×××××.com |