1 |
On Mon, 15 Aug 2011 09:09:04 +0200 |
2 |
Ulrich Mueller <ulm@g.o> wrote: |
3 |
> >> I'd still suggest that for existing EAPIs we should go with |
4 |
> >> mgorny's latest patch, simply because "packages common to DEPEND |
5 |
> >> and RDEPEND (but see below)" is a more accurate description than |
6 |
> >> "none". |
7 |
> |
8 |
> > But if we do that, people will claim it's the package mangler's |
9 |
> > fault if packages common to DEPEND and RDEPEND aren't available. |
10 |
> > You're asking for something unimplementable to be specified. |
11 |
> |
12 |
> No, they can't claim that, because we explicitly say that there are |
13 |
> restrictions in case of circular dependencies. |
14 |
|
15 |
The problem is, there are nearly always circular dependencies. Any |
16 |
package that depends upon autotools is part of a hundred-plus-large |
17 |
circular dependency cycle. So if we're saying packages can count on |
18 |
RDEPENDs being merged beforehand except when necessary to resolve |
19 |
cycles, then we need to say how cycles are to be resolved. |
20 |
|
21 |
-- |
22 |
Ciaran McCreesh |