1 |
2010-12-18 03:48:26 Donnie Berkholz napisaĆ(a): |
2 |
> On 02:45 Sat 18 Dec , Arfrever Frehtes Taifersar Arahesis wrote: |
3 |
> > Problem #1: USE flags cannot contain "." characters. |
4 |
> |
5 |
> This isn't a problem, it's an arbitrary statement of an antifeature. My |
6 |
> understanding of the actual problem here is that you want to have some |
7 |
> sort of USE flags for various Python versions with names containing |
8 |
> periods, for usability. Perhaps you could expand on exactly what you |
9 |
> want to do, so we can work together to figure out whether this is the |
10 |
> best route to solve your problem. |
11 |
> |
12 |
> If the problem is handling which Python versions to build modules for, |
13 |
> I'm wondering whether enhancing the eselect module to select multiple |
14 |
> preferred versions might not be a better solution than EAPI changes. |
15 |
|
16 |
USE flags would allow to use USE dependencies to ensure that dependencies of given package |
17 |
have been installed for required Python versions. |
18 |
|
19 |
> > ================================================================================================ |
20 |
> > |
21 |
> > Problem #2: Files in profiles cannot use features from newer EAPIs. |
22 |
> |
23 |
> Could you explain how the eapi file in profiles does not address this? |
24 |
> The 10.0 profiles are using it already with EAPI=2. |
25 |
|
26 |
I would like to use EAPI="4"-specific syntax in a file in base profile (or somewhere else) to |
27 |
affect all profiles. If I have to rely on "eapi" files, then all non-deprecated profiles |
28 |
checked by repoman would have to use EAPI >=4. |
29 |
|
30 |
> > ================================================================================================ |
31 |
> > |
32 |
> > Problem #3: repoman doesn't allow stable packages to have optional dependencies on unstable |
33 |
> > packages (usually until these packages are stabilized). |
34 |
> |
35 |
> This seems useful at first glance, but I'm wondering how big of a |
36 |
> problem it really is and whether this solution is a bit overarchitected. |
37 |
> |
38 |
> > Example of the problem: |
39 |
> > If "python_abis_2.7", "python_abis_3.1" and "python_abis_3.2" USE |
40 |
> > flags are masked using use.mask on given architectures until Python |
41 |
> > 2.7, 3.1 and 3.2 are stabilized on these architectures, then majority |
42 |
> > of reverse dependencies of Python won't be tested with new versions of |
43 |
> > Python. |
44 |
> |
45 |
> Why does that have to be the case? Why not unmask them earlier but only |
46 |
> make them available to ~arch ebuilds? |
47 |
|
48 |
There are already hundreds of stable ebuilds, which support unstable Python 2.7 (without |
49 |
explicit optional dependency on Python 2.7). The solution to problem #3 shouldn't require |
50 |
any changes in ebuilds. |
51 |
|
52 |
> I don't know of anyone who's actually done this, but setting IUSE based |
53 |
> on ACCEPT_KEYWORDS being ~arch should be possible. |
54 |
|
55 |
It would break invariance of metadata. |
56 |
|
57 |
-- |
58 |
Arfrever Frehtes Taifersar Arahesis |