1 |
On 2011-05-10, Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
2 |
> Apparently, though unproven, at 16:40 on Tuesday 10 May 2011, Grant Edwards |
3 |
> did opine thusly: |
4 |
> |
5 |
>> I ran emerge --depclean the other day on one of my machines and it |
6 |
>> removed Python 2.6. I was using Python 2.6 as my "default" python, |
7 |
>> and depclean's removal of it broke a _lot_ of stuff. About a half |
8 |
>> day's worth of hassle later I had Python 2.6 re-installed and my |
9 |
>> system was again usable. |
10 |
>> |
11 |
>> In order to avoid the same circus on my other machines, how do I |
12 |
>> prevent emerge --depclean from removing Python 2.6? |
13 |
> |
14 |
> Put that slot in world: |
15 |
> |
16 |
>=dev-lang/python:2.6 |
17 |
> |
18 |
> I suppose there are better and more automagically elegant ways of doing it, |
19 |
> but this works. |
20 |
|
21 |
Thanks! |
22 |
|
23 |
(you need to leave out the '='). |
24 |
|
25 |
> I think the issue happens because portage does not take eselect |
26 |
> choices into account when building it's dep graph, it only uses the |
27 |
> DEPENDS in ebuilds. |
28 |
|
29 |
Apparently so. It seems like it ought to pay attention to eselect. |
30 |
If I've explicitly configured my system to use 2.6 instead of 2.7, |
31 |
removing 2.6 doesn't seem like a good thing... |
32 |
|
33 |
> You likely have nothing left that explicitly uses 2.6 and all the |
34 |
> ebuilds depend only on python 2 point something |
35 |
> |
36 |
> When you finally choose to remove python-2.6, you simply have to |
37 |
> emerge -C it and not rely on --depclean |
38 |
|
39 |
Yup. |
40 |
|
41 |
-- |
42 |
Grant Edwards grant.b.edwards Yow! I like the way ONLY |
43 |
at their mouths move ... They |
44 |
gmail.com look like DYING OYSTERS |