| 1 |
On Sat, 13 Aug 2011 10:38:28 +0200 |
| 2 |
Ulrich Mueller <ulm@g.o> wrote: |
| 3 |
> >>>>> On Sat, 13 Aug 2011, Ciaran McCreesh wrote: |
| 4 |
> > I think we're going about this in the wrong way. We should probably |
| 5 |
> > remove all mention of circular dependencies, and just state that |
| 6 |
> > there's nothing except system things guaranteed for pkg_*. |
| 7 |
> |
| 8 |
> Except that large parts of the tree rely on packages in RDEPEND being |
| 9 |
> available in pkg_*. |
| 10 |
|
| 11 |
Then those packages are broken. (And we can't ban RDEPEND cycle |
| 12 |
breaking, since large parts of the tree rely upon that too.) |
| 13 |
|
| 14 |
> > If there's a need for dependencies that will definitely be installed |
| 15 |
> > for pkg_setup, we should introduce an IDEPEND (for 'install'). |
| 16 |
> |
| 17 |
> And then at some point we will have circular IDEPEND dependencies and |
| 18 |
> the package manager will have to break such cycles, as it does for |
| 19 |
> RDEPEND now. |
| 20 |
|
| 21 |
No, IDEPEND will simply not allow cycle breaking, in the same way as |
| 22 |
DEPEND. IDEPEND would be, in effect, DEPEND that is also honoured for |
| 23 |
binary packages. |
| 24 |
|
| 25 |
-- |
| 26 |
Ciaran McCreesh |
Archives