1 |
On Sat, 7 Nov 2015 00:12:56 +0000 |
2 |
David Leverton <levertond@××××××××××.com> wrote: |
3 |
|
4 |
> Ulrich Mueller wrote: |
5 |
> >>>>>> On Fri, 6 Nov 2015, David Leverton wrote: |
6 |
> >> One last thing (probably): this should also be |
7 |
> >> included/cross-referenced in tree-layout.tex for the repo-level |
8 |
> >> package.mask file. |
9 |
> > |
10 |
> > This hasn't been discussed so far, and I think it is way too late to |
11 |
> > consider it still for EAPI 6. |
12 |
> |
13 |
> Well, a big use case stated in bug 282296 is package.mask in overlays, |
14 |
> which I assume refers to the repo-level one rather than one inside an |
15 |
> actual profile. If people haven't made it clear that they want that, I |
16 |
> think it's because of imprecise language rather than because they don't |
17 |
> actually want it. |
18 |
> |
19 |
> (And this has been argued over for long enough - I'd really rather not |
20 |
> have it go on ever further when people realise the spec still doesn't |
21 |
> allow what they're trying to do.) |
22 |
> |
23 |
> > Also, why should package.mask be treated differently from the other |
24 |
> > files in the top-level profiles/ directory? |
25 |
> |
26 |
> For consistency, they should probably all (except repo_name and eapi) |
27 |
> allow directories, yes, but a couple of the others were mentioned in the |
28 |
> bug and it seems people decided there wasn't any need to allow it for |
29 |
> the rest of them. Going the other way and making none of them allow |
30 |
> directories isn't really better in that regard because then the repo |
31 |
> package.mask is different from the profile one. |
32 |
|
33 |
I'm pretty sure that was *the* request, just the convoluted tree layout |
34 |
caused it to be misunderstood. Then, I'd vote for leaving it out until |
35 |
EAPI 7. |
36 |
|
37 |
-- |
38 |
Best regards, |
39 |
Michał Górny |
40 |
<http://dev.gentoo.org/~mgorny/> |