1 |
On Mon, 7 Dec 2015 15:36:30 -0500 |
2 |
Mike Gilbert <floppym@g.o> wrote: |
3 |
|
4 |
> On Sun, Dec 6, 2015 at 2:03 PM, Michał Górny <mgorny@g.o> wrote: |
5 |
> > Hi, |
6 |
> > |
7 |
> > Here's another patchset. Mostly fixups inspired by multilib-build.eclass |
8 |
> > changes with a few little additions. |
9 |
> > |
10 |
> > Changes: |
11 |
> > |
12 |
> > 1. eclass-set variables are now read-only, |
13 |
> |
14 |
> Sounds ok, but might break some ebuilds. |
15 |
> |
16 |
> > 2. 'unset -f' is used to unset temporary & local functions, |
17 |
> > 3. implementations are reordered for sane order. |
18 |
> > |
19 |
> > I've tested this with a few dozen random distutils-r1, python-r1, |
20 |
> > python-any-r1 and python-single-r1 packages. However, for |
21 |
> > the implementation reorder a larger tinderbox run would be appreciated. |
22 |
> > |
23 |
> > As explained in the commit, the reorder may influence files installed by |
24 |
> > a package, and implementation selected by python_setup(). This should |
25 |
> > not cause issues for correctly written ebuilds, and should help us find |
26 |
> > those that are not correctly written ;-). |
27 |
> > |
28 |
> > In other words, we're finally considering Python 3.x preferred over |
29 |
> > Python 2.x. |
30 |
> |
31 |
> That could be a significant change, and I do expect that this will |
32 |
> break some ebuilds. |
33 |
> |
34 |
> Any easy way to test this first? Or should we just be ready for the bug reports? |
35 |
|
36 |
Not an easy. We could try some tinderboxing with USE='doc python'. Or |
37 |
just grep ebuilds for python_export_best calls and plain python_setup |
38 |
calls, and see if someone didn't assume it will always prefer python2. |
39 |
REQUIRED_USE with python_gen_useflags may be a suggestion too. |
40 |
|
41 |
So I'd go for waiting for bug reports. Or maybe doing the change |
42 |
conditionally to EAPI 6 for improved confusion ;-). |
43 |
|
44 |
-- |
45 |
Best regards, |
46 |
Michał Górny |
47 |
<http://dev.gentoo.org/~mgorny/> |