1 |
Ciaran McCreesh wrote: |
2 |
> On Mon, 16 Oct 2006 00:25:42 +0200 Jakub Moc <jakub@g.o> wrote: |
3 |
> | Ciaran McCreesh napsal(a): |
4 |
> | > The profiles change over time. Currently, when the profiles change, |
5 |
> | > the only thing that has to be checked for conflicting USE behaviour |
6 |
> | > is subprofiles. With IUSE defaults, the person making the change |
7 |
> | > will also have to do a sanity check over the entire tree. |
8 |
> | |
9 |
> | Uh, what kind of conflicting behaviour and what sanity checks are you |
10 |
> | talking about here? Did you _really_ miss the whole point of this |
11 |
> | feature? |
12 |
> |
13 |
> Before changing default values for USE flags, arch and release people |
14 |
> have to make sure that that change won't do something nasty like |
15 |
> introduce circular or built_with_use deps into the default system |
16 |
> resolution. |
17 |
> |
18 |
|
19 |
I don't see how the location of the default USE affects these things. |
20 |
If I change default USE in my ebuild; I have to do sanity checks. If I |
21 |
change default USE in the profile; I have to do sanity checks *in that |
22 |
profile*. |
23 |
|
24 |
So if your argument is that it's cheaper to check just N profiles ( the |
25 |
profiles affected by my change ) versus all available profiles; then I |
26 |
agree with you on that point. |
27 |
|
28 |
However I still believe there exist examples where default USE in an |
29 |
ebuild is a better solution. |
30 |
-- |
31 |
gentoo-dev@g.o mailing list |