1 |
On Thu, Mar 10, 2016 at 7:44 AM, Alexis Ballier <aballier@g.o> wrote: |
2 |
> |
3 |
> The above examples are needed in order to be able to provide working stuff, |
4 |
> predate PMS and do not conform to it. The only issue they cause is that |
5 |
> alternative PMs might not implement them properly. |
6 |
> |
7 |
|
8 |
I think the intent is to get stuff like this into PMS or change it, |
9 |
not to just start breaking things arbitrarily. |
10 |
|
11 |
The underlying message is that devs shouldn't just quietly rely on |
12 |
non-PMS behavior without putting it on a tracker and use breakage as |
13 |
an excuse. If we're depending on non-PMS behavior we need to get that |
14 |
onto a tracker so that either PMS or the packages can be fixed as |
15 |
appropriate, and then once the issue is dealt with we need to stick to |
16 |
PMS. |
17 |
|
18 |
In any case, these are my general thoughts on the issue. When |
19 |
packages are still not aligned with PMS we should be tracking it and |
20 |
fixing either the package or PMS. Of course, we could spot some |
21 |
non-PMS-compliant behavior that predates PMS 10 years from now, and it |
22 |
isn't like we can just have QA just comment out some ebuild lines |
23 |
without any regard to what it does to the tree. When PMS issues come |
24 |
up yesterday, tomorrow, or 10 years from now we have to deal with |
25 |
their reality. That said, we still need to DEAL with them and not |
26 |
just ignore them. |
27 |
|
28 |
-- |
29 |
Rich |