1 |
On sob, 2017-05-20 at 15:30 +0200, Michał Górny wrote: |
2 |
> Hi, everyone. |
3 |
> |
4 |
> Here's a set of patches inspired by the recent Sphinx dependency |
5 |
> discussion. They make python-r1 (and therefore distutils-r1) capable |
6 |
> of any-of dependency logic similar to the one used in python-any-r1. |
7 |
> |
8 |
> The basic goal is relatively simple -- to improve handling of pure |
9 |
> build-time dependencies in the eclass. It solves two common problems: |
10 |
> |
11 |
> a. dependencies on packages that support only a subset of PYTHON_COMPAT, |
12 |
> |
13 |
> b. dependencies that need to be implementation-bound between themselves |
14 |
> (e.g. Sphinx plugins). |
15 |
> |
16 |
> The new API improves both of those cases significantly. For the former, |
17 |
> we no longer force user to select additional targets via REQUIRED_USE -- |
18 |
> instead, we just any-of dependencies + python_check_deps() to select |
19 |
> implementation independently of whether it is enabled or not. |
20 |
> |
21 |
> For the latter, we no longer have to force all targets of the package |
22 |
> on all the involved dependencies. Again, using any-of dep |
23 |
> and appropriate python_check_deps() we can enforce a single (any) |
24 |
> target throughout all the packages and use it. |
25 |
> |
26 |
> The first three patches do some code refactoring that makes the change |
27 |
> easier and possibly improves maintainability of the code. The next two |
28 |
> patches add support for python_check_deps() and python_gen_any_dep() |
29 |
> respectively. The last two patches provide examples for both use cases |
30 |
> mentioned. |
31 |
> |
32 |
|
33 |
Merged them a while back. |
34 |
|
35 |
-- |
36 |
Best regards, |
37 |
Michał Górny |