| 1 |
On 8/15/15 7:35 AM, Michał Górny wrote: |
| 2 |
> Dnia 2015-08-15, o godz. 00:05:57 |
| 3 |
> Johannes Huber <johu@g.o> napisał(a): |
| 4 |
>> if we want to attract more contributors we should consider to have one |
| 5 |
>> supported EAPI (latest). EAPI 4 is the last not marked as deprecated |
| 6 |
>> (< EAPI 5). The move in ebuilds from EAPI 4 to EAPI 5 is very simple |
| 7 |
>> replacing the declaration. As we have now git in place we could easily |
| 8 |
>> do this with one commit (awesome!!!!). |
| 9 |
> |
| 10 |
> This is a cheap hack, not a conversion. Proper conversion to a new EAPI |
| 11 |
> is about using the new EAPI features. Not marking it 'done', |
| 12 |
> and pretending there's nothing more to do. |
| 13 |
|
| 14 |
Good point. |
| 15 |
|
| 16 |
My understanding here it's more about deprecating EAPI 4 which we |
| 17 |
wouldn't be able to do without moving to a newer EAPI. |
| 18 |
|
| 19 |
Also note that even when writing from scratch, it's not guaranteed that |
| 20 |
the developer will use all relevant EAPI 5 features. |
| 21 |
|
| 22 |
Nothing seems to prevent doing the mass conversion first, deprecating |
| 23 |
EAPI 4, and then having a second, slower pass to make sure the ebuilds |
| 24 |
are using EAPI 5 as they should be. One way might be to add a "TODO" |
| 25 |
comment during mass conversion, which then can be grepped for (and |
| 26 |
removed after manual review). |
| 27 |
|
| 28 |
Paweł |
Archives