* Mike Frysinger vap...@gentoo.org schrieb:
although portage has long been generating the NEEDED files in vdb. even
stable portage generates these files.
Ah, okay, I wasn't aware of that.
What's the difference between NEEEDED and NEEDED.2 ? Multiarch ?
cu
--
* Micha?? Górny mgo...@gentoo.org schrieb:
What do you think about this idea ?
You mean what do we think about portage-2.2 and preserved-libs?
Well, I'm still using portage-2.1, so I wans't aware of whats going
on there. For now it seems the preservation is still done explicitly
On Saturday, January 01, 2011 23:09:11 Enrico Weigelt wrote:
BTW: several blog/maillist postings talked about the problem that
even on recompile, older library versions could be linked in even
on recompile.
you'll need to provide an actual example. i have yet to see one.
-mike
signature.asc
On 01/01/2011 08:09 PM, Enrico Weigelt wrote:
Well, I'm still using portage-2.1, so I wans't aware of whats going
on there. For now it seems the preservation is still done explicitly
(preserve_old_lib calls in certain ebuilds ?). My proposal is to
record the necessary information (eg. which so
On Sunday, January 02, 2011 00:17:03 Zac Medico wrote:
On 01/01/2011 08:09 PM, Enrico Weigelt wrote:
Well, I'm still using portage-2.1, so I wans't aware of whats going
on there. For now it seems the preservation is still done explicitly
(preserve_old_lib calls in certain ebuilds ?). My
Hi folks,
just a little braindump, how revdep-rebuild could be made
(partially) obsolete in future:
Today it might happen that on an library update an old .so file
gets unmerged while still used by somebody - that's what revdep-
rebuild scans for. While it should catch those cases, we still
On 12/31/2010 03:42 PM, Enrico Weigelt wrote:
What do you think about this idea ?
I think you should check out the preserve-libs feature in portage-2.2.
On Fri, 31 Dec 2010 21:42:32 +0100
Enrico Weigelt weig...@metux.de wrote:
The main problem IMHO is that portage doesn't record which libraries
some package links in, so it doesn't know which ones have to be
protected from unmerge (unless explicitly stated somewhere).
So I'd propose to add
...
Subsequent merges will update this that,
...
Subsequent merges may happen after a long while. Old, possibly
vulunerable library will still be used, and most likely unseen by
admin.
