1 |
On Wed, 28 Nov 2012 13:25:25 +0100 |
2 |
Dirkjan Ochtman <djc@g.o> wrote: |
3 |
|
4 |
> Patches 1 and 2 look good to me. Patch 3 looks like it merits some |
5 |
> higher-level design discussion? |
6 |
|
7 |
Yes, that's why I sent it here. |
8 |
|
9 |
Shortly saying, the issue is that some packages (like chromium) need |
10 |
Python for build-time only but don't install anything related to it. |
11 |
|
12 |
Giving user the choice of implementation is not really necessary / |
13 |
relevant there. Therefore, the idea is to provide a minimalistic code |
14 |
to handle it with least effort possible. |
15 |
|
16 |
In the simple case, chromium does: |
17 |
|
18 |
PYTHON_COMPAT=( python{2_6,2_7} ) |
19 |
inherit python-build-r1 |
20 |
|
21 |
DEPEND=${PYTHON_DEPS} |
22 |
|
23 |
#pkg_setup() { |
24 |
# python-build-r1_pkg_setup |
25 |
# ... |
26 |
#} |
27 |
|
28 |
and the build completes gracefully. |
29 |
|
30 |
And the whole effort is necessary to make sure that the active Python |
31 |
implementation is one supported by package. For example, the above |
32 |
ebuild would use (or even might pull in) python2 if python3 is active. |
33 |
|
34 |
-- |
35 |
Best regards, |
36 |
Michał Górny |