1 |
Hello, |
2 |
|
3 |
As some of you are aware already, there is a significant difference |
4 |
in handling Python implementation choice between python.eclass |
5 |
and python-r1 / p-d-ng. |
6 |
|
7 |
The python.eclass packages use USE_PYTHON (undocumented variable) |
8 |
or the eselected Python2+3, and store that choice implicitly. |
9 |
Everything works on the assumption that implementations don't change |
10 |
or user runs python-updater. |
11 |
|
12 |
The python-r1 / p-d-ng packages use explicit USE flags to choose |
13 |
implementation. The choice is thus explicitly exported and other |
14 |
packages can use USE dependencies to enforce matching implementations. |
15 |
|
16 |
The problems arise whenever python-r1 package depends on package using |
17 |
the python.eclass, or the other way round. There is no clear way to |
18 |
enforce the same implementations being used, so everything falls down |
19 |
to user. |
20 |
|
21 |
In order to solve the most common problems due to that, I have |
22 |
published a news item explaining how PYTHON_TARGETS work and what needs |
23 |
to be done with them. |
24 |
|
25 |
In the discussion with other Gentoo Python devs, it was assumed that |
26 |
USE_PYTHON should not be considered public and thus not be described |
27 |
in the message. Sadly, it seems that this resulted in people actually |
28 |
trying to disable Python3 using PYTHON_TARGETS and seeing random |
29 |
failures due to python.eclass enforcing it anyway. |
30 |
|
31 |
|
32 |
I believe we should thoroughly think what should be done now in order |
33 |
to solve the issues best for the transition time. Here are a few |
34 |
solutions I can think of. |
35 |
|
36 |
1) lard both eclasses with USE_PYTHON <-> PYTHON_TARGETS consistency |
37 |
checks. This should be pretty easy for python-r1, harder for |
38 |
python.eclass (anyone volunteering to touch it in thick gloves?). |
39 |
|
40 |
In other words, we just warn users whenever they need to do something |
41 |
to ensure things work. |
42 |
|
43 |
|
44 |
2) make python.eclass respect PYTHON_TARGETS. |
45 |
|
46 |
Sadly, due to PMS limitations we can't tell if PYTHON_TARGETS was set |
47 |
by user or has the default value. So we either can: |
48 |
|
49 |
a) stop using USE_PYTHON and eselect defaults, start using |
50 |
PYTHON_TARGETS only, pull in correct Python versions, tell users to run |
51 |
python-updater; |
52 |
|
53 |
b) prefer PYTHON_TARGETS over USE_PYTHON and the default (but only |
54 |
if PYTHON_TARGETS != default PYTHON_TARGETS, so we're on thin ice here). |
55 |
|
56 |
|
57 |
3) make python.eclass respect and export PYTHON_TARGETS in IUSE. |
58 |
|
59 |
Like the above but instead of running the fragile and broken |
60 |
python-updater, we simply force rebuild of all packages. People can |
61 |
start using USE-deps in the meantime, everyone becomes a bit happier |
62 |
in the further plane. Sadly, this means every Python package is going |
63 |
to need being rebuilt. |
64 |
|
65 |
|
66 |
Please also note that 2 & 3 mean mass-changing behavior of stable |
67 |
packages. Not something I'd do in a short timeframe where the issue |
68 |
should be best fixed. |
69 |
|
70 |
Any other ideas? Any +1, -1s? |
71 |
|
72 |
-- |
73 |
Best regards, |
74 |
Michał Górny |