| 1 |
On 12/18/19 11:34 AM, Ulrich Mueller wrote: |
| 2 |
> |
| 3 |
> Removal of the virtual/emacs ebuilds won't remove the installed package |
| 4 |
> from users' systems. It will eventually disappear, when all its reverse |
| 5 |
> dependencies have been updated. Why would its continued presence as an |
| 6 |
> installed package (for another while) cause any problems? |
| 7 |
|
| 8 |
Unless the VDB is updated, portage will see a dependency on a package |
| 9 |
that doesn't exist and could refuse to do a lot of things like a @world |
| 10 |
update involving a rebuild of one of those packages, or a --depclean. |
| 11 |
|
| 12 |
This *does* happen if you mask virtual/emacs. It *could* happen if you |
| 13 |
delete it. |
| 14 |
|
| 15 |
|
| 16 |
> Revbumping its more than 400 reverse dependencies really doesn't sound |
| 17 |
> so attractive, and would cause rebuilds on users' systems for virtually |
| 18 |
> (pun intended :-) no benefit. |
| 19 |
|
| 20 |
If portage bails on an update and I have to troubleshoot the problem for |
| 21 |
ten minutes, then that's already wasted more of my time than if it |
| 22 |
reinstalled all 400 revdeps. The benefit is that people don't get |
| 23 |
cryptic messages from a confused packaged manager that they have to |
| 24 |
debug all day. |
Archives