1 |
On Wed, 31 Oct 2012 13:07:17 -0400 |
2 |
Mike Gilbert <floppym@g.o> wrote: |
3 |
|
4 |
> On Wed, Oct 31, 2012 at 5:17 AM, Michał Górny <mgorny@g.o> wrote: |
5 |
> > 3) add an '#if' clause and build the Python-relevant code conditionally. |
6 |
> > |
7 |
> > In other words, let python-exec install both regular 'python-exec' |
8 |
> > and the 'python' binary built from the same code with '#if' switched. |
9 |
> > |
10 |
> > Pros: |
11 |
> > |
12 |
> > - still simple, and clean. |
13 |
> > |
14 |
> > Cons: |
15 |
> > |
16 |
> > - two executables installed. |
17 |
> > |
18 |
> > |
19 |
> > What are your thoughts? |
20 |
> > |
21 |
> |
22 |
> If others agree that replacing python-wrapper is a good idea to begin |
23 |
> with, this option would be my choice. |
24 |
|
25 |
Well, I don't see much of a point to have two wrappers which serve |
26 |
almost the same purpose with the main difference being that one does it |
27 |
the simple almost-dumb way, and the other implementing a piece of |
28 |
package manager inside. |
29 |
|
30 |
> FYI, bypassing the setenv(GENTOO_PYTHON_PROCESS_NAME, ...) that is set |
31 |
> in python-wrapper.c could cause issues with python-based daemons |
32 |
> called from init scripts. If a pidfile is not used, openrc might not |
33 |
> be able to find the right process. However, a pidfile seems to be a |
34 |
> better solution for that anyway. |
35 |
> |
36 |
> I'm sure djc would love to drop that patch from our dev-lang/python patchset. |
37 |
|
38 |
I'd love to get rid of all that GENTOO_* stuff as well. If they serve |
39 |
a real purpose, that means we're doing something just horribly wrong. |
40 |
|
41 |
In any case, do I understand correctly that python-exec does break all |
42 |
those packages anyway? We may simply decide to wait till it gets more |
43 |
wide-spread (and thus packages get fixed to work with it) before |
44 |
replacing python-wrapper. |
45 |
|
46 |
-- |
47 |
Best regards, |
48 |
Michał Górny |