1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA256 |
3 |
|
4 |
On 15/08/15 02:48 AM, Andrew Savchenko wrote: |
5 |
> On Sat, 15 Aug 2015 08:12:42 +0200 Paweł Hajdan, Jr. wrote: |
6 |
>> On 8/15/15 3:16 AM, Ian Stakenvicius wrote: |
7 |
>>> Secondly, though, conversion to EAPI5 is not actually |
8 |
>>> trivial, there are a couple of things, 'usex' related for |
9 |
>>> instance, that also need to be taken care of. If it was just |
10 |
>>> a matter of running a sed -e 's/^EAPI=4/EAPI=5/' on all |
11 |
>>> in-tree ebuilds this would have been done a long time ago. |
12 |
>> |
13 |
>> Do you mean that it would break things or just while |
14 |
>> technically working the ebuilds mass converted in this way |
15 |
>> would not take advantage of EAPI5 features such use 'usex'? |
16 |
> |
17 |
> Another issue: EAPI change normally requires revision bump. |
18 |
> Revision bump for stable packages will require stabilization |
19 |
> request in most cases (e.g. if all current stable versions are |
20 |
> <EAPI 5). Stabilization request often take ages. |
21 |
|
22 |
I don't think that one's actually true in the general case. |
23 |
|
24 |
EAPI5 conversions needed revbumps because the addition of subslots |
25 |
or slot operators to dependencies requires a revbump (and EAPI5 adds |
26 |
default :${SLOT}/${SLOT} to all deps in VDB for portage at least) |
27 |
|
28 |
That said, it is another reason in this particular case, thanks! |
29 |
-----BEGIN PGP SIGNATURE----- |
30 |
Version: GnuPG v2 |
31 |
|
32 |
iF4EAREIAAYFAlXPOyUACgkQAJxUfCtlWe1BwgD/feoCRfMWQdzj0lpoHDBr2HJz |
33 |
AcuEi7WWsxPyjn4FSqYBALQNPyjOTbr7B6T41J3pA/wgdbwTSZJr01un2GoyqZ7y |
34 |
=1qRi |
35 |
-----END PGP SIGNATURE----- |