1 |
On Sun, 11 Nov 2012 14:29:20 +0800 |
2 |
Ben de Groot <yngwin@g.o> wrote: |
3 |
|
4 |
> On 11 November 2012 00:43, Michał Górny <mgorny@g.o> wrote: |
5 |
> |
6 |
> > Hello, |
7 |
> > |
8 |
> > As some of you are aware already, there is a significant difference |
9 |
> > in handling Python implementation choice between python.eclass |
10 |
> > and python-r1 / p-d-ng. |
11 |
> > |
12 |
> > The python.eclass packages use USE_PYTHON (undocumented variable) |
13 |
> > or the eselected Python2+3, and store that choice implicitly. |
14 |
> > Everything works on the assumption that implementations don't change |
15 |
> > or user runs python-updater. |
16 |
> > |
17 |
> > The python-r1 / p-d-ng packages use explicit USE flags to choose |
18 |
> > implementation. The choice is thus explicitly exported and other |
19 |
> > packages can use USE dependencies to enforce matching implementations. |
20 |
> > |
21 |
> > The problems arise whenever python-r1 package depends on package using |
22 |
> > the python.eclass, or the other way round. There is no clear way to |
23 |
> > enforce the same implementations being used, so everything falls down |
24 |
> > to user. |
25 |
> > |
26 |
> > In order to solve the most common problems due to that, I have |
27 |
> > published a news item explaining how PYTHON_TARGETS work and what needs |
28 |
> > to be done with them. |
29 |
> > |
30 |
> > In the discussion with other Gentoo Python devs, it was assumed that |
31 |
> > USE_PYTHON should not be considered public and thus not be described |
32 |
> > in the message. Sadly, it seems that this resulted in people actually |
33 |
> > trying to disable Python3 using PYTHON_TARGETS and seeing random |
34 |
> > failures due to python.eclass enforcing it anyway. |
35 |
> > |
36 |
> > |
37 |
> > I believe we should thoroughly think what should be done now in order |
38 |
> > to solve the issues best for the transition time. Here are a few |
39 |
> > solutions I can think of. |
40 |
> > |
41 |
> > 1) lard both eclasses with USE_PYTHON <-> PYTHON_TARGETS consistency |
42 |
> > checks. This should be pretty easy for python-r1, harder for |
43 |
> > python.eclass (anyone volunteering to touch it in thick gloves?). |
44 |
> > |
45 |
> > In other words, we just warn users whenever they need to do something |
46 |
> > to ensure things work. |
47 |
> > |
48 |
> > |
49 |
> > 2) make python.eclass respect PYTHON_TARGETS. |
50 |
> > |
51 |
> > Sadly, due to PMS limitations we can't tell if PYTHON_TARGETS was set |
52 |
> > by user or has the default value. So we either can: |
53 |
> > |
54 |
> > a) stop using USE_PYTHON and eselect defaults, start using |
55 |
> > PYTHON_TARGETS only, pull in correct Python versions, tell users to run |
56 |
> > python-updater; |
57 |
> > |
58 |
> > b) prefer PYTHON_TARGETS over USE_PYTHON and the default (but only |
59 |
> > if PYTHON_TARGETS != default PYTHON_TARGETS, so we're on thin ice here). |
60 |
> > |
61 |
> > |
62 |
> > 3) make python.eclass respect and export PYTHON_TARGETS in IUSE. |
63 |
> > |
64 |
> > Like the above but instead of running the fragile and broken |
65 |
> > python-updater, we simply force rebuild of all packages. People can |
66 |
> > start using USE-deps in the meantime, everyone becomes a bit happier |
67 |
> > in the further plane. Sadly, this means every Python package is going |
68 |
> > to need being rebuilt. |
69 |
> > |
70 |
> > |
71 |
> > Please also note that 2 & 3 mean mass-changing behavior of stable |
72 |
> > packages. Not something I'd do in a short timeframe where the issue |
73 |
> > should be best fixed. |
74 |
> > |
75 |
> > Any other ideas? Any +1, -1s? |
76 |
> I'm sorry if I missed something, but I don't see why USE_PYTHON should be |
77 |
> considered not public. That has been the only mechanism until now (in |
78 |
> combination with package.mask) to get a consistent system without redundant |
79 |
> multiple python versions. |
80 |
|
81 |
Well, the most important reason is that I don't think USE_PYTHON is |
82 |
pretty much documented *anywhere*. |
83 |
|
84 |
> I think it is easiest to tell users to set USE_PYTHON as well, until |
85 |
> python.eclass is finally phased out. |
86 |
|
87 |
Yes, that's what the warning would say. |
88 |
|
89 |
-- |
90 |
Best regards, |
91 |
Michał Górny |