1 |
Jakub Moc wrote: |
2 |
> Alec Warner napsal(a): |
3 |
>> Jakub Moc wrote: |
4 |
>>> Danny van Dyk napsal(a): |
5 |
>>>> which breaks the metadata cache. Any objections to change it |
6 |
>>>> to |
7 |
>>>> |
8 |
>>>> SLOT=0 |
9 |
>>> As noted on the relevant bug [1], the eclass is a complete no-op and |
10 |
>>> nothing can be installed using this eclass (has been so for quite some |
11 |
>>> time). Fixing it doesn't make sense, making it dummy or even removing it |
12 |
>>> (plus the unusable single ebuild which inherits it) does. |
13 |
>>> |
14 |
>>> |
15 |
>>> [1] http://bugs.gentoo.org/show_bug.cgi?id=162960 |
16 |
>>> |
17 |
>> And we already told you, removing it isn't a good solution (even if it |
18 |
>> technically works within the bounds of the current api). The bug is |
19 |
>> that the SLOT invalidates cache entries and it's trivial to fix. |
20 |
> |
21 |
> The eclass is not trivial to fix, to be fixed, this eclass would require |
22 |
> a complete rewrite from scratch. There's no usable ebuild for this |
23 |
> eclass and the eclass is completely moot. Still fail to see what are you |
24 |
> trying to fix as opposed to removing a broken non-functional cruft from |
25 |
> the tree. |
26 |
> |
27 |
> |
28 |
|
29 |
See the bug? It says 'slot is being set to $KV, which breaks the |
30 |
metadata cache. Also, portage may fail to set $KV to a valid slot value |
31 |
(typically 0) and thus the package may end up with SLOT="" which is also |
32 |
invalid'. |
33 |
|
34 |
Thats what we are trying to fix. |
35 |
|
36 |
The fact that the eclass sucks balls is irrelevant, I could say the same |
37 |
of a dozen other old as hell eclasses. But they remain in the tree, as |
38 |
will this one. |
39 |
-- |
40 |
gentoo-dev@g.o mailing list |