1 |
July 20, 2018 2:55 PM, "Rich Freeman" <rich0@g.o> wrote: |
2 |
|
3 |
> On Fri, Jul 20, 2018 at 8:39 AM <nado@××××××××××.be> wrote: |
4 |
> |
5 |
>> Why not introducing a new level in the hierarchy ? Something like "common" could be fit. |
6 |
>> |
7 |
>> default/linux/amd64/13.0 |
8 |
>> default/linux/amd64/13.0/common |
9 |
>> default/linux/amd64/13.0/common/desktop |
10 |
>> default/linux/amd64/13.0/common/developer |
11 |
>> ... |
12 |
>> |
13 |
>> By doing so we could still have a bare profiles with minimal things set to work, and have the |
14 |
>> common subset with sane defaults for most users. |
15 |
> |
16 |
> I think one of the issues is that our docs/defaults and the mentality |
17 |
> of our users tends to drive them to what looks like the most basic |
18 |
> starting point. |
19 |
> |
20 |
> I think that having a base profile intended just as an inheritance |
21 |
> point for other profiles makes sense technically, but it may not |
22 |
> actually be a good default for end-users. |
23 |
> |
24 |
> If you set up the example above, how many would would still pick 13.0 |
25 |
> as their starting point, and not common? |
26 |
> |
27 |
> Now, there is nothing that says that inheritance has to follow the |
28 |
> directory tree. We could have a |
29 |
> default/linux/amd64/13.0/donotuse/core profile that everything |
30 |
> inherits, and make that the minimal one. It just means that most |
31 |
> profiles under 13.0 wouldn't inherit 13.0. |
32 |
> |
33 |
> To some degree we may have painted ourselves into a bit of a corner by |
34 |
> presenting this as a heirarchy, as it tends to force the bottom of the |
35 |
> heirarchy to be the best profile for inheritance, when it is also the |
36 |
> first thing users see as well. |
37 |
|
38 |
I’m not sure I was clear enough in what 13.0 would mean : basically, its current content would be |
39 |
delegated to common, and 13.0 would keep only things needed to have minimal breakages/conflicts. |
40 |
And we would keep the current directory-like inheritance. |
41 |
|
42 |
Then about what is presented to users and how they choose it : we could adapt eselect-profile to |
43 |
hilight some suggested defaults by use-case and people who still want to pick the minimalist one |
44 |
can do so freely, as they are probably doing it right now anyway. |
45 |
|
46 |
-- |
47 |
Corentin “Nado” Pazdera |