1 |
Am 03.08.2011 23:44, schrieb Willie Wong: |
2 |
> On Sun, Jul 31, 2011 at 01:38:58PM +0200, Alan McKinnon wrote: |
3 |
>> It's sensible really - portage is not the only package manager out |
4 |
>> there and therefore should not be in @system. The user did not put |
5 |
>> portage in world, and did not use -D, so portage is not updating the |
6 |
>> package. |
7 |
>> |
8 |
>> The solution is simple - all users should put their preferred package |
9 |
>> manager into world and what Stroller is seeing will stop happening. |
10 |
>> |
11 |
>> Zac can't force portage into system like he could with less and nano |
12 |
>> and have few or non side-effects. A virtual package manager only says |
13 |
>> that you *have* one, not *which* one. So as usual for Gentoo, the user |
14 |
>> gets to tell the software which one it is. |
15 |
>> |
16 |
>> I don't see a problem. |
17 |
> |
18 |
> Though it is silly IMHO that portage would want to remove itself with |
19 |
> depclean. Could it not be hardcoded into portage that it should try to |
20 |
> keep itself updated and not commit suicide? (Independently of the |
21 |
> @system sets.) |
22 |
> |
23 |
> W |
24 |
|
25 |
I don't really see an issue here. There are lots of packages whose |
26 |
removal will wreak havok on your system: wget, gcc, python, binutils |
27 |
etc. Some are part of @system and are therefore protected. For others |
28 |
like portage there are alternatives which means that AFAIK they cannot |
29 |
be part of @system. None of these have any protection except that |
30 |
dependencies will usually prevent their removal by emerge --depclean. |
31 |
|
32 |
For portage itself, Albert already pointed out that it ought to be |
33 |
protected from --depclean. |
34 |
|
35 |
Portage doesn't protect you from shooting yourself in the foot with |
36 |
`emerge -C foo`. It just tries not to it by itself when you ask it |
37 |
kindly with `emerge -c foo`. ;) |
38 |
|
39 |
Regards, |
40 |
Florian Philipp |