1 |
On 30 Aug 2008, at 13:56, Alan McKinnon wrote: |
2 |
|
3 |
> The main reason these packages are behind at all is that they are |
4 |
> usually |
5 |
> build dependencies, not run dependencies. They will only be updated |
6 |
> with |
7 |
> emerge -uD when something that depends on them is rebuilt. |
8 |
> |
9 |
> To avoid this, use 'emerge --with-bdeps y' |
10 |
> This has the side effect of knowing what to do with SLOTs |
11 |
|
12 |
So a periodic 'emerge --with-bdeps world' would be worthwhile? |
13 |
|
14 |
> In general I find that emerge is infinitely better at knowing how |
15 |
> to get what |
16 |
> I want than I am, so it's always best to let it do what it wants to do |
17 |
|
18 |
I'd really debate this premise. Perhaps the problem is not with |
19 |
`emerge` itself, perhaps with the ebuilds or with simple versioning |
20 |
incompatibilities, but the number of cock-ups one sees with emerged |
21 |
packages... well, I think "infinitely" good is stretching it just a |
22 |
little. |
23 |
|
24 |
I'm not saying Portage is poor - other package managers have given me |
25 |
more headaches per usage. Maybe the problem is with build-time |
26 |
dependencies of the build-time dependencies, I don't know, but when I |
27 |
had the libexpat.so.0 error the only thing that worked (having |
28 |
followed a number of different advices posted here) was to rebuild |
29 |
EVERY outdated package on my system - a total numbering in the region |
30 |
of 250. I wouldn't have imagined I had so many packages installed, |
31 |
never mind those "missed" by my regular updates. |
32 |
|
33 |
Stroller. |