1 |
Robert Coie wrote: |
2 |
> Is there a good reason for handling package.mask differently from the |
3 |
> various profiles in /usr/portage/profiles? IOW, would it be a problem |
4 |
> to have portage look at /etc/package.mask (for example), which would |
5 |
> be a symlink to one of several choices in /usr/portage/profiles? This |
6 |
> would seem to facilitate separate package masks for different |
7 |
> architectures, and would allow machines of different architectures to |
8 |
> more easily share a locally mirrored portage tree. |
9 |
|
10 |
package.mask actually serves a different purpose than the packages file. |
11 |
package.mask is for blocking packages from all profiles, usually |
12 |
because they are unstable, untested, or just plain broken -- in all |
13 |
profiles. The packages file is a fine-grained specification of what |
14 |
files are *acceptable* in a profile. The packages file could take the |
15 |
place of the package.mask, but, as Spider pointed out, that would mean |
16 |
making entries in all the profiles' packages files to mask out broken |
17 |
packages. Furthermore, Every user can have their own profile (I |
18 |
recommend it for the most flexibility in your setup). In that case, |
19 |
package.mask is really necessary to make sure a broken package is not |
20 |
merged accidentally. Think of it this way: package.mask is our |
21 |
(Gentoo's) list of packages that *cannot* be installed, even if they are |
22 |
in the portage tree. The packages file is your list of files that *can* |
23 |
be installed. |
24 |
|
25 |
Chad (chadh@g.o) |