1 |
On Sun, 07 Sep 2008 17:24:55 +0400 |
2 |
Peter Volkov <pva@g.o> wrote: |
3 |
> В Вск, 07/09/2008 в 02:05 +0000, Jorge Manuel B. S. Vicetto пишет: |
4 |
> > Our first attempt was to use a multislot use flag[1]. According to |
5 |
> > that flag, we would set the SLOT and the PREFIX for the install. |
6 |
> > That has the a very important problem - it breaks the invariancy of |
7 |
> > the SLOT and as thus been put aside. |
8 |
> |
9 |
> Could you explain in a little bit more details why it's bad? How |
10 |
> portage workarounds this now for binutils? |
11 |
|
12 |
Portage uses the metadata cached slot when doing dep resolution, so it |
13 |
goes spectacularly wrong. |
14 |
|
15 |
> In any case as FHS and /usr/kde/<version> installations should set |
16 |
> differently SLOT seems that new portage feature is required... May be |
17 |
> portage should allow setting SLOT in dependence on USE flag? |
18 |
|
19 |
So what's the slot when SLOT="foo? ( 1 ) bar? ( 2 )"? And should you be |
20 |
able to have the same KDE version with both USE=multislot and |
21 |
USE=-multislot installed at the same time? |
22 |
|
23 |
Unfortunately, the issue's not as simple as allowing conditionals in |
24 |
SLOT in a future EAPI. |
25 |
|
26 |
> Or new property for PROPERTIES called multislot which will workaround |
27 |
> the problem with "invariancy of the SLOT"? |
28 |
|
29 |
Uh, how would that work? |
30 |
|
31 |
-- |
32 |
Ciaran McCreesh |