1 |
Hi, |
2 |
|
3 |
A portage question, to satisfy my own curiosity i.e. I know how to fix |
4 |
the problem I had, but I'd like to know if I fully understand how to |
5 |
use portage in a case like this: |
6 |
|
7 |
My desktop is a mostly stable x86 machine, but I use e17-9999 as my wm. |
8 |
x11-wm/e depends on evas which has a hard dependency on cairo: |
9 |
|
10 |
cairo? ( >=x11-libs/cairo-1.2 ) |
11 |
|
12 |
I had to unmask e, evas and cairo to get this to emerge. |
13 |
|
14 |
I also have x86 pygtk and pycairo already installed, pycairo has a hard |
15 |
dependency on cairo-1.0* |
16 |
|
17 |
=x11-libs/cairo-1.0* |
18 |
|
19 |
emerge world wants to downgrade cairo to 1.0* to satisfy pycairo, then |
20 |
the next emerge world wants to upgrade it to 1.2* to satisfy evas = a |
21 |
version conflict which yo-yos back and forth. The fix is obviously to |
22 |
keyword pycairo to ~x86 to give consistent dependencies. This is harder |
23 |
than it could be: pygtk could also have had it's own conflicts, and it |
24 |
takes several 'emerge -pv world' and reading ebuilds to get the whole |
25 |
picture. And this is something portage could have figured out itself. |
26 |
|
27 |
emerge -p world had to parse the entire portage tree anyway, it could |
28 |
easily check for versioning conflicts while doing it and display a |
29 |
message to the user, much like blockers do currently. I get it that |
30 |
portage likely can't suggest resolutions but it would be nice to see in |
31 |
one display that evas and pygtk requires mutually exclusive versions of |
32 |
cairo. |
33 |
|
34 |
Can portage do this? Is it something useful that warrants a feature |
35 |
request? Or am I missing an option in the man page even though I've |
36 |
read it >30 times? :-) |
37 |
|
38 |
alan |
39 |
-- |
40 |
gentoo-user@g.o mailing list |