1 |
This is something I've thought about too and I agree that this would be |
2 |
a useful feature. One possible application of this feature is to |
3 |
simplify how we handle simd extensions on amd64 (this has been discussed |
4 |
multiple times). Currently we have the USE flags mmx,sse,3dnow masked on |
5 |
amd64 because then enable x86 specific assembly in a lot of packages. |
6 |
Those packages that work on amd64 with the relevent simd extension we |
7 |
hard-enable on amd64. This has several problems: |
8 |
|
9 |
1. Users cannot easily disable these features for e.g debugging |
10 |
purposes. |
11 |
2. It's not clear to the user that these extensions are being enabled as |
12 |
emerge -pv shows (-mmx) etc. |
13 |
|
14 |
Now if we could simply mask the use flags for the packages where it |
15 |
causes problems and not for the ones that don't we could solve both |
16 |
these issues. |
17 |
|
18 |
Herbie. |
19 |
|
20 |
On Mon, 2005-06-13 at 20:43 -0600, Jason Wever wrote: |
21 |
> On Mon, 13 Jun 2005 16:40:48 +0200 |
22 |
> Sven Wegener <swegener@g.o> wrote: |
23 |
> |
24 |
> > We just had a short discussion over in #gentoo-portage and the idea of |
25 |
> > an use.force file for profiles came up. |
26 |
> |
27 |
> One feature that would be more useful (in my honest on Tuesdays |
28 |
> opinion) for us arch folks is the ability to mask use flags on a |
29 |
> per-package basis. Often times use flags will work for 99% of the |
30 |
> packages they are used in, but the other 1% will not. Currently the |
31 |
> workaround is to just make the ebuild ignore that use flag on that arch, |
32 |
> but there's no real indication to the user that the workaround is |
33 |
> thwarting their use flag preferences (unless the arch monkey is nice |
34 |
> enough to put in some einfo love). |
35 |
> |
36 |
> Cheers, |