1 |
Ciaran McCreesh wrote: |
2 |
> And why does repoman do that? |
3 |
> |
4 |
> Oh. Yeah. Because people with an attitude like yours think that the |
5 |
> correct way to fix a repoman message is to start nuking arch keywords, |
6 |
> ignoring what it does to the rest of the tree. |
7 |
|
8 |
Dropping keywords works perfectly to have repoman quit complaining, you |
9 |
just have to do a recursive dropping on the rdeps of this package. |
10 |
|
11 |
> Perhaps because the people maintaining those archs have better things |
12 |
> to do that deal with the same silly ill-thought-out arguments every |
13 |
> three months. |
14 |
|
15 |
cia/cvs commits ml says something different, gentoo wise at least. |
16 |
|
17 |
>> I mean, if vapier can maintain arm/sh/s390, by himself, to a better |
18 |
>> degree than the mips *TEAM* can do, that should be an indication of a |
19 |
>> problem. |
20 |
> |
21 |
> That's an interesting assertion. Can you back it up? |
22 |
|
23 |
Feel free to run imlate scripts and come up with some numbers. |
24 |
|
25 |
Note that I hate whining and I love get solutions. |
26 |
|
27 |
MOST of the packages runs fine if they build fine, MOST of the |
28 |
endian-issues or the 64bit-issues got caught by ppc and amd64 and there |
29 |
aren't that many right now. Ugly arch specific codepath could be |
30 |
present, but, as I said, usually you catch those breaking on gcc. So |
31 |
having some way to test if the package builds (cross toolchain) and if |
32 |
the package at least runs (qemu) IS something that should let small |
33 |
arches with large tree coverage improve a bit. Otherwise you can just |
34 |
reduce the tree coverage. |
35 |
|
36 |
lu |
37 |
|
38 |
-- |
39 |
|
40 |
Luca Barbato |
41 |
Gentoo Council Member |
42 |
Gentoo/linux Gentoo/PPC |
43 |
http://dev.gentoo.org/~lu_zero |
44 |
|
45 |
-- |
46 |
gentoo-dev@l.g.o mailing list |