1 |
On Thu, 1 Apr 2010 21:58:29 +0300 |
2 |
Dror Levin <spatz@g.o> wrote: |
3 |
> On Thu, Apr 1, 2010 at 20:32, Ciaran McCreesh < |
4 |
> ciaran.mccreesh@××××××××××.com> wrote: |
5 |
> > In that case, why would you like to see VALID_USE as well as |
6 |
> > pkg_pretend? Why not just use pkg_pretend, which is already in EAPI |
7 |
> > 4 and which can do everything VALID_USE can do plus several useful |
8 |
> > things that it can't? |
9 |
> > |
10 |
> > With pkg_pretend you can do VALID_USE using a simple eclass... |
11 |
> > |
12 |
> With pkg_pretend you can also implement USE deps, would you like to |
13 |
> abolish them as well? |
14 |
|
15 |
Use dependencies can be handled by the package manager in a useful way. |
16 |
|
17 |
> This is data and should be treated as such. |
18 |
> Using code to describe it is possible but suboptimal, if there's a |
19 |
> suggestion that allows for describing this as data rather than with |
20 |
> code then it is superior for all the reasons already stated. |
21 |
|
22 |
There's nothing that can make use of it as data, so there's no point. |
23 |
|
24 |
If, in the distant future, something becomes available that can make |
25 |
use of it as data, then in a future EAPI we can start using both as |
26 |
appropriate. |
27 |
|
28 |
-- |
29 |
Ciaran McCreesh |