1 |
Am Samstag, den 31.03.2012, 14:44 +0200 schrieb Ulrich Mueller: |
2 |
> >>>>> On Sat, 31 Mar 2012, Maciej Grela wrote: |
3 |
> |
4 |
> > I've read the PMS and I haven't found information whether this variable |
5 |
> > is supposed to be set during pkg_prepare or not. |
6 |
> |
7 |
> There is no such stage. You mean pkg_pretend, I suppose? |
8 |
> |
9 |
> > Therefore I ask, what is the proper behaviour here ? Is there |
10 |
> > documentation on what special env variables are supposed to be |
11 |
> > defined in each stage ? |
12 |
> |
13 |
> It's specified here: |
14 |
> <http://dev.gentoo.org/~ulm/pms/4/pms.html#x1-11900012.1> |
15 |
> |
16 |
> | Variable Legal in Consistent? Description |
17 |
> | --------------------------------------------------------------------- |
18 |
> | T All Partially⁴ The full path to a temporary |
19 |
> | directory for use by the ebuild. |
20 |
> | |
21 |
> | ⁴Consistent and preserved across a single connected sequence of |
22 |
> | install or uninstall phases, but not between install and uninstall. |
23 |
> | When reinstalling a package, this variable must have different |
24 |
> | values for the install and the replacement. |
25 |
> |
26 |
> > Can this be considered as a bug in paludis ? |
27 |
> |
28 |
> The spec seems to be clear that T is legal in all phases, including |
29 |
> pkg_pretend. |
30 |
|
31 |
Well, I'd say: there is no sane value you can assign to $T since you are |
32 |
not allowed to write anything anyway: |
33 |
|
34 |
"pkg_pretend must not write to the |
35 |
filesystem." (http://dev.gentoo.org/~ulm/pms/4/pms.html#x1-9700010.1.2) |
36 |
|
37 |
and since "pkg_pretend is run separately from the main phase function |
38 |
sequence, and does not participate in any kind of environment saving" it |
39 |
is not guaranteed to be set to the same $T later. |
40 |
|
41 |
Cheers, |
42 |
Tiziano |