| 1 |
Ian Stakenvicius <axs@g.o> wrote: |
| 2 |
> |
| 3 |
> Keeping every single dependency around and valid just so that pkg_*rm |
| 4 |
> can completely cleanly seems like so much overkill, though.. |
| 5 |
|
| 6 |
It is not only overkill, it would require a merging strategy which |
| 7 |
AFAIK portage currently does not use and which would lead to blockers: |
| 8 |
|
| 9 |
If you really want to guarantee that all := dependencies are |
| 10 |
satisfied in the moment of unmerging, you need to do the following. |
| 11 |
|
| 12 |
Assume foo depends on bar:=. Now if bar gets an upgrade, first |
| 13 |
foo must be unmerged, then bar must be upgraded, and then |
| 14 |
foo must be reemerged. |
| 15 |
|
| 16 |
If you have a circular dependency of := chains, an upgrade |
| 17 |
is impossible forever: You would manually have to break this |
| 18 |
chain by setting some useflags exactly in the same way as |
| 19 |
you did when you manages to install the chain for the first time. |
| 20 |
Otherwise, you will just miss the possibly avilable updates |
| 21 |
forever. |
| 22 |
|
| 23 |
The problems if you really want to guarantee that all |
| 24 |
dependencies are still installed when unmerging a package |
| 25 |
are hardly reasonably solvable. |
| 26 |
|
| 27 |
One should better make a rule that if an ebuild has such a |
| 28 |
special requirement that it needs a certain version for |
| 29 |
being unmerged, it must block all other versions |
| 30 |
(one could make a syntax for the := case or even disallow |
| 31 |
this case for such ebuilds) instead of letting the PM jump |
| 32 |
through hoops for the "normal" cases. |
Archives