1 |
On 2022-08-31 17:27, Mike Gilbert wrote:
|
2 |
|
3 |
> The only pain point I see is for users with /usr on a separate |
4 |
> filesystem and that are not using an appropriate initramfs. |
5 |
|
6 |
As mentioned on IRC earlier on today, another (although this would be
|
7 |
less of a pain point and more of a sneaky changes to run-time behaviour
|
8 |
of a system) might be ebuilds which rely on PATH precedence of /usr/bin
|
9 |
over /bin to override binaries installed in the latter by other
|
10 |
packages. Some examples:
|
11 |
|
12 |
- app-arch/lbzip2[symlink] - installs /usr/bin/b{,un}zip2 to override
|
13 |
/bin/b{,un}zip2 installed by app-arch/bzip2;
|
14 |
|
15 |
- app-arch/pigz[symlink] - ditto for g{,un}zip and app-arch/gzip;
|
16 |
|
17 |
- app-arch/gzip installs the symlink /bin/uncompress pointing to
|
18 |
gunzip, app-arch/ncompress installs /usr/bin/uncompress pointing to
|
19 |
compress.
|
20 |
|
21 |
There probably are more though, and I feel these will all need a
|
22 |
systemic change - eselect modules? shell aliases? - to how such
|
23 |
overrides are handled. Still, for the time being packages which have
|
24 |
such behaviour controlled by USE flags should me added to
|
25 |
package.use.force of relevant profiles.
|
26 |
|
27 |
--
|
28 |
Marecki |