| 1 |
Dnia 2015-08-15, o godz. 09:06:48 |
| 2 |
Ulrich Mueller <ulm@g.o> napisał(a): |
| 3 |
|
| 4 |
> >>>>> On Sat, 15 Aug 2015, Paweł Hajdan, wrote: |
| 5 |
> |
| 6 |
> > Nothing seems to prevent doing the mass conversion first, |
| 7 |
> > deprecating EAPI 4, and then having a second, slower pass to make |
| 8 |
> > sure the ebuilds are using EAPI 5 as they should be. One way might |
| 9 |
> > be to add a "TODO" comment during mass conversion, which then can be |
| 10 |
> > grepped for (and removed after manual review). |
| 11 |
> |
| 12 |
> How about a line like "EAPI=4" which can be grepped for? :) |
| 13 |
> Seriously, what would be the benefit of a conversion if you need a |
| 14 |
> second pass to redo it properly? |
| 15 |
> |
| 16 |
> Also there are still some 5000 ebuilds (14 %) with EAPIs 0, 1, 2, and |
| 17 |
> 3 in the tree. IMHO it would be better to spend our effort in updating |
| 18 |
> those first. |
| 19 |
|
| 20 |
First, we'd use a tool which would tell us precisely which old ebuild |
| 21 |
versions can really be removed. Like 'eshowkw' but also grepping other |
| 22 |
ebuilds for dependencies that can not be satisfied by newer versions. |
| 23 |
|
| 24 |
|
| 25 |
-- |
| 26 |
Best regards, |
| 27 |
Michał Górny |
| 28 |
<http://dev.gentoo.org/~mgorny/> |
Archives