1 |
On Sun, 2006-10-15 at 22:01 +0100, Ciaran McCreesh wrote: |
2 |
> On Sun, 15 Oct 2006 22:35:10 +0200 Jakub Moc <jakub@g.o> wrote: |
3 |
> | > Which is why I suggested changing Portage's behaviour earlier in the |
4 |
> | > thread. Like it or not, overlays are already getting complex enough |
5 |
> | > that they'd benefit from profile behaviour. |
6 |
> | |
7 |
> | Because maintaining your own profiles and stacking them and dealing |
8 |
> | with all the related mess is a _lot_ easier that sticking a + before |
9 |
> | foo in IUSE. Right. ;) |
10 |
> |
11 |
> You mean, than sticking a + before foo in IUSE in every ebuild, and |
12 |
> ensuring that changes are kept in sync and consistent with the |
13 |
> behaviour of every single existing profile. |
14 |
|
15 |
No one here is talking about doing that... |
16 |
|
17 |
What we are talking about is an instance where foo is *not* enabled by |
18 |
default in profiles but there is *one* package where it is upstreams |
19 |
intention that foo be enabled by default but they still provide the |
20 |
capability to turn foo support off. This package (and all of the ebuilds |
21 |
that are in the tree for it) would have a +foo in IUSE...thus even |
22 |
though foo is generally off unless the user specifies -foo in either |
23 |
make.conf or package.use foo is turned on for this package and this |
24 |
package alone. |
25 |
|
26 |
No one is talking about replacing tree wide defaults with this |
27 |
functionality...this is for package maintainers to specify default |
28 |
behavior for their package and their package alone independent of the |
29 |
profiles intent. |
30 |
|
31 |
Doing it your way in order to make sure that a package was built the way |
32 |
a maintainer intended (by default) they would have to make an entry in |
33 |
package.use in every single tier one profile (at the moment only |
34 |
base)...this is also something that they cannot enforce over external |
35 |
overlays...so it looses any value at all. |
36 |
|
37 |
--Dan |