1 |
On Thursday 25 September 2008 17:51:42 Hilco Wijbenga wrote: |
2 |
> On Thu, Sep 25, 2008 at 03:27, Daniel Pielmeier |
3 |
> |
4 |
> <daniel.pielmeier@××××××××××.com> wrote: |
5 |
> > Also make sure you have a clean world file, which means if there is an |
6 |
> > entry which is already needed by something else it should normally be |
7 |
> > removed as this could cause problems. |
8 |
> |
9 |
> I can understand that if I have both package P and dependency D in |
10 |
> world, and I remove P then depclean will not remove D. I assume that's |
11 |
> not what you're talking about, though, as it's harmless and probably |
12 |
> intended (since I listed D in world). |
13 |
> |
14 |
> Moreover, I would think this is quite common. You might have |
15 |
> Java/Perl/Python/Ruby/* explicitely listed in world (because you write |
16 |
> programs/scripts) and several packages that are either |
17 |
> Java/Perl/Python/Ruby/* based or provide bindings. I *want* to be |
18 |
> explicit here as I don't want Java/Perl/Python/Ruby/* removed should I |
19 |
> remove those other packages. |
20 |
> |
21 |
> So what kind of problems are we talking about? |
22 |
|
23 |
Most probably the "problem" that arises is when a mere dep cannot be removed |
24 |
by --depclean anymore because it is in world. It is a problem because of the |
25 |
unnecessary clutter it creates. |
26 |
|
27 |
Also, sometimes deps come and go, and libs get split into two packages. Say |
28 |
you have libA which is a dep of some package Z. The dev sees that libA can |
29 |
benefit from USE flags in some way and this in turn benefits package Z. So |
30 |
libA gets split into libAa and libAb, both of which block the old libA. |
31 |
Package Z can install libAa or libAb or both depending on USE. If the user |
32 |
put the old libA in world (incorrectly), portage will not uninstall it to |
33 |
replace it with the new split libA. Instead it will display the blockers, |
34 |
which has to be resolved manually - and we all know how much mass hysteria |
35 |
and confusion blockers seem to cause amongst users... |
36 |
|
37 |
This of course is the exact opposite of what you describe above |
38 |
|
39 |
-- |
40 |
alan dot mckinnon at gmail dot com |