1 |
On Sat, 19 Apr 2008 18:38:06 +0200 |
2 |
"Marijn Schouten (hkBst)" <hkBst@g.o> wrote: |
3 |
> Every package dependency in DEPEND is installed and usable before |
4 |
> src_unpack starts, right? So is the question here whether or not they |
5 |
> can be uninstalled right before pkg_{pre,post}inst starts? |
6 |
|
7 |
If we're using binaries, DEPEND is usually ignored. |
8 |
|
9 |
> I don't know what the general use of pkg_preinst is, but in |
10 |
> pkg_postinst the package itself should be runnable, so its RDEPENDS |
11 |
> should be installed and usable at this point. So perhaps we should |
12 |
> define that "usable" means "each of its RDEPENDs is installed and has |
13 |
> had its pkg_postinst function run". The recursion of that definition |
14 |
> then comes from the requirement that RDEPENDs should be usable before |
15 |
> pkg_postinst starts running. |
16 |
|
17 |
No good. That prevents RDEPEND <-> RDEPEND cycles from being solved, |
18 |
and the package manager has to be able to solve that. |
19 |
|
20 |
> If only one of those packages has a pkg_postinst then it is still |
21 |
> solvable. If they both have a pkg_postinst then one of those is |
22 |
> probably not essential for the actual usability of the package and |
23 |
> should be removed. A final possibility is that the pkg_postinsts are |
24 |
> both necessary for a fully functional package but not for the |
25 |
> functionality used in the other pkg_postinst. If this is the case, |
26 |
> then perhaps we should specify deps according to which ebuild phase |
27 |
> they are necessary for? |
28 |
|
29 |
Not with current EAPIs we can't. |
30 |
|
31 |
> SRC_UNPACK_DEP="app-arch/unzip" |
32 |
> SRC_COMPILE_DEP="dev-scheme/bigloo" |
33 |
> SRC_INSTALL_DEP="" |
34 |
|
35 |
Labels are a cleaner solution to this. But again, we're discussing |
36 |
current EAPIs here. |
37 |
|
38 |
-- |
39 |
Ciaran McCreesh |