| 1 |
On Tuesday 28 April 2009 01:20:56 Michael Higgins wrote: |
| 2 |
> I _remember_ doing this just a few days ago, so I'm not doing it again... |
| 3 |
> but why is it still throwing a warning? Is this a limitation of portage or |
| 4 |
> the ebuild? |
| 5 |
> |
| 6 |
> -------------- |
| 7 |
> WARN: postinst |
| 8 |
> /usr/lib/cups exists - You need to remerge every ebuild that |
| 9 |
> installed into /usr/lib/cups and /etc/cups, qfile is in portage-utils: |
| 10 |
> # FEATURES=-collision-protect emerge -va1 $(qfile -qC /usr/lib/cups |
| 11 |
> /etc/cups | sed "s:net-print/cups$::") FEATURES=-collision-protect is |
| 12 |
> needed to overwrite the compatibility symlinks installed by this package, |
| 13 |
> it won't be needed on later merges. |
| 14 |
|
| 15 |
There's the "solution" to your problem. Every time you rebuild cups, it will |
| 16 |
check and find that this directory exists, and assume that it's an upgrade |
| 17 |
that will break packages linking to it. You could enable -collision-protect, |
| 18 |
but I can't honestly recommend that. |
| 19 |
|
| 20 |
> You should also run revdep-rebuild |
| 21 |
> -------------- |
| 22 |
> |
| 23 |
> Did they forget to tell me what to do so that the warning stops? '-) |
| 24 |
|
| 25 |
No, I think the initial check in pkg_postinst() is a little too wide. Merely |
| 26 |
checking for the existence of a directory will give false positives. It would |
| 27 |
be better to check for the existence of particular files and symlinks so the |
| 28 |
ebuild knows more exactly what is going on. |
| 29 |
|
| 30 |
I'd file a bug, and ask for this check to be made more specific |
| 31 |
|
| 32 |
> Is it possible that it's necessary to re-emerge all those packages on every |
| 33 |
> cups *ebuild* update? |
| 34 |
|
| 35 |
Did you upgrade or just rebuild? With an upgrade, I'd revdep-rebuild and with |
| 36 |
a rebuild, you could probably ignore this warning. |
| 37 |
|
| 38 |
|
| 39 |
-- |
| 40 |
alan dot mckinnon at gmail dot com |
Archives