1 |
On Sun, 18 Jan 2009 13:21:55 +0200 |
2 |
Petteri Räty <betelgeuse@g.o> wrote: |
3 |
> One thing to note why it's bad to rely on it is that if you have an |
4 |
> eclass setting RDEPEND then you are probably not getting what you |
5 |
> wanted. |
6 |
|
7 |
Actually, you do. If you have ebuild: |
8 |
|
9 |
DEPEND="from/ebuild" |
10 |
|
11 |
and eclass: |
12 |
|
13 |
DEPEND="from/eclass" |
14 |
RDEPEND="also-from/eclass" |
15 |
|
16 |
you end up with: |
17 |
|
18 |
DEPEND="from/ebuild from/eclass" |
19 |
RDEPEND="from/ebuild also-from/eclass" |
20 |
|
21 |
It's been that way for several years now. |
22 |
|
23 |
And yes, I'd really like to see this killed for EAPI 3. Ideally we'd go |
24 |
with a single DEPENDENCIES variable with labels of some kind, |
25 |
something like: |
26 |
|
27 |
DEPENDENCIES=" |
28 |
build: |
29 |
foo/bar |
30 |
build+run: |
31 |
foo/baz |
32 |
post: |
33 |
foo/plugin" |
34 |
|
35 |
which would make it much easier to start specifying dependencies for |
36 |
use of ROOT properly in the future -- it'd just be new labels, not |
37 |
zillions of new variables. |
38 |
|
39 |
-- |
40 |
Ciaran McCreesh |