| 1 |
On Tue, 24 Feb 2009 10:46:30 -0500 |
| 2 |
Richard Freeman <rich0@g.o> wrote: |
| 3 |
> I will certainly concede that putting it inside the ebuild |
| 4 |
> potentially breaks compatibility with existing package managers. |
| 5 |
> That is certainly a downside to this approach. However, none of the |
| 6 |
> other objections that have been raised appear to hold water. |
| 7 |
|
| 8 |
...and it means we can't change name or version rules. |
| 9 |
|
| 10 |
...and it means over doubling the best possible time to work out a |
| 11 |
dependency tree in the common case where the metadata cache is valid. |
| 12 |
|
| 13 |
...and it means we can't make arbitrary format changes. |
| 14 |
|
| 15 |
> And if backwards compatibility were a serious issue you could define |
| 16 |
> a new ".ebuild2" file spec that incorporates the EAPI inside the file |
| 17 |
> and current package managers would ignore it. Then you're not |
| 18 |
> changing the file extension every time a new EAPI comes along, and |
| 19 |
> the need to do so could be handled via future GLEPs. |
| 20 |
|
| 21 |
Developers already have to stop and think and consult the conveniently |
| 22 |
available table of features for EAPIs. By splitting the EAPI concept in |
| 23 |
two you're doubling the amount of data to be learnt. |
| 24 |
|
| 25 |
-- |
| 26 |
Ciaran McCreesh |
Archives