1 |
* Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
2 |
> On Tuesday 18 March 2008, Enrico Weigelt wrote: |
3 |
> > BTW: if libexpat.so.0 was there before the update and now isn't, |
4 |
> > there's an major bug in expat (either the ebuild or the source). |
5 |
> |
6 |
> There's no bug in expat, the OP is doing an *expat*upgrade*, which |
7 |
> means that libexpat.so.0 was there before and libexpat.so.1 is there |
8 |
> now. Thus any app that links explicitly to libexpat.so.0 is now broken. |
9 |
|
10 |
If an upgrade breaks primary things, it's simply broken. |
11 |
You can now choose, whether the bug is expat or the ebuild/portage. |
12 |
|
13 |
> 2. Something like revdep-rebuild that scans the system looking for |
14 |
> errors and is triggered manually by the user. |
15 |
|
16 |
This is also ugly and error-prone. After every slightest upgrade |
17 |
you have to fear nothing's running anymore and everything has to be |
18 |
rebuilt first. If downtimes don't matter, fine. But for critical |
19 |
systems, this is very dangerous. |
20 |
|
21 |
Why not modeling/storing the *runtime* dependencies and doing |
22 |
an *clean* MVCC ? Isn't this what slotting was meant for ? ;-P |
23 |
|
24 |
> Imagine using a binary distro where the packager didn't |
25 |
> pick this up and you are left stranded with no easy way to fix it... |
26 |
|
27 |
Then the package manager's obviously broken ... |
28 |
|
29 |
|
30 |
cu |
31 |
-- |
32 |
--------------------------------------------------------------------- |
33 |
Enrico Weigelt == metux IT service - http://www.metux.de/ |
34 |
--------------------------------------------------------------------- |
35 |
Please visit the OpenSource QM Taskforce: |
36 |
http://wiki.metux.de/public/OpenSource_QM_Taskforce |
37 |
Patches / Fixes for a lot dozens of packages in dozens of versions: |
38 |
http://patches.metux.de/ |
39 |
--------------------------------------------------------------------- |
40 |
-- |
41 |
gentoo-user@l.g.o mailing list |