| 1 |
Dne neděle 22 Únor 2009 22:34:19 |
| 2 |
> Carsten Lohrke wrote: |
| 3 |
> On Freitag, 20. Februar 2009, Petteri Räty wrote: |
| 4 |
> > Suggestions/objections? |
| 5 |
> |
| 6 |
> If you mean by "mass migration" doing that more or less blindly, I do |
| 7 |
> object. When an ebuild directly or indirectly inherits an eclass, which is |
| 8 |
> EAPI 2 enabled, like base.eclass, while another isn't, you have to expect |
| 9 |
> side-effects. See for example media-tv/kdetv-0.8.9-r1, which features an |
| 10 |
> empty src_prepare to prevent the attempt to apply patches twice, |
| 11 |
> temporarily. |
| 12 |
> |
| 13 |
> So the first step is to get all eclasses EAPI 2 ready and even then I |
| 14 |
> wouldn't rule out odd cases, so changes should happen in testing and |
| 15 |
> revised ebuilds exclusively to assure odd cases get caught. |
| 16 |
> |
| 17 |
> |
| 18 |
> Carsten |
| 19 |
Well that is the reason why i am first eapi2ing the kde eclass. I was really |
| 20 |
suprised when i saw kde3 ebuilds with eapi2 :( |
| 21 |
The smalest problems i can spot in the first place are: |
| 22 |
1) not working PATCHES array |
| 23 |
2) double configure script run. |
| 24 |
|
| 25 |
So i agree that we have to doublecheck every eclass and eapi2fy them in first |
| 26 |
place. |
| 27 |
|
| 28 |
Tomas |
Archives