1 |
On 25-03-2009 16:44:59 +0100, Michael Haubenwallner wrote: |
2 |
> If you mean having prefix-keywords in KEYWORDS, then no: live-ebuilds |
3 |
> might be prefix ready or not, and they usually just have empty KEYWORDS. |
4 |
> To mark an ebuild prefix ready, IMO it should set PREFIX_KEYWORDS. |
5 |
|
6 |
What's the difference? ~x86 != ~x86-linux. Ebuilds with empty KEYWORDS |
7 |
technically say they don't work for anything. It remains a problem, |
8 |
even when PREFIX_KEYWORDS is set, simply because I don't think you can |
9 |
(and should) be able to tell if a variable is set in order to mask it in |
10 |
the resolving phase. You just evaluate somehow which keywords are to be |
11 |
accepted, and then you figure out it is masked. |
12 |
|
13 |
> > For haubi-style ACCEPT_KEYWORDS we need a GLEP/proposal on gentoo-dev |
14 |
> > for the next EAPI. |
15 |
> |
16 |
> Agreed, although I don't think this should be an EAPI thing, as EAPI-0 |
17 |
> ebuilds can have distro-keywords too. |
18 |
|
19 |
That's true, but you need metadata cache for this not to be awfully |
20 |
slow, so it won't be painless. Hence I prefer to stick with my |
21 |
observation for now that |
22 |
( gentoo-x86-keywords ) U ( prefix-keyword ) = Ø |
23 |
So, in what what way you want to administer them, they remain the same. |
24 |
|
25 |
|
26 |
-- |
27 |
Fabian Groffen |
28 |
Gentoo on a different level |