| 1 |
>>>>> On Mon, 05 Jul 2021, Aaron Bauman wrote: |
| 2 |
|
| 3 |
> On Mon, Jul 05, 2021 at 12:17:35PM +0200, Ulrich Mueller wrote: |
| 4 |
>> In the past, the council had also banned EAPIs. However, that didn't |
| 5 |
>> make much of a practical difference because an EAPI cannot be added |
| 6 |
>> to eapis-banned in layout.conf unless all ebuilds are gone. Maybe we |
| 7 |
>> should have a rule like the following instead: |
| 8 |
>> "A deprecated EAPI is considered banned when the Gentoo repository |
| 9 |
>> no longer contains any ebuilds using it." |
| 10 |
|
| 11 |
> Having the council vote to officially ban an EAPI does have |
| 12 |
> practicality. The work done by those porting ebuilds, removing old |
| 13 |
> packages, etc is hindered when other begin committing ebuilds |
| 14 |
> deprecated but not banned. |
| 15 |
|
| 16 |
> I would ask that the council continue the official motions and votes |
| 17 |
> to ban EAPI's prior to being "enforced" by tooling. This assists those |
| 18 |
> doing the work and the QA team to stop people from committing. |
| 19 |
|
| 20 |
The point is that banning an EAPI doesn't have any noticeable effect. |
| 21 |
For example, if you look at EAPI 0 (banned on 2016-01-10) and EAPI 4 |
| 22 |
(banned on 2018-04-08), there's neither a cusp nor any change of slope |
| 23 |
visible for the curves plotted here: |
| 24 |
https://www.akhuettel.de/~huettel/plots/eapi.php |
| 25 |
|
| 26 |
Ulrich |
Archives