1 |
On 29 Jun, Dale wrote: |
2 |
> Helmut Jarausch wrote: |
3 |
>> Hi, |
4 |
>> |
5 |
>> I'm using portage-2.2_rc33. And it has become smarter with every new |
6 |
>> release. Still, I don't understand why it's sometimes smarter than other |
7 |
>> times. |
8 |
>> |
9 |
>> In most cases now, it's able to handle cases where installed old |
10 |
>> packages which depend on, say, libXX-version-1, block emerging |
11 |
>> libXX-version-2 by unmerging the old version itself. |
12 |
>> |
13 |
>> But sometimes, like today when upgrading from various qt-...-4.5.1 |
14 |
>> to qt-...-4.5.2 packages it produces tons of blockings which it |
15 |
>> cannot resolve itself. |
16 |
>> |
17 |
>> How can I help portage to resolve this blockings itself in a smart way? |
18 |
>> (trying to do |
19 |
>> emerge --keep-going -j2 -1 --ask --update --newuse --deep @system @world |
20 |
>> automatically. |
21 |
>> ) |
22 |
>> |
23 |
>> Many thanks for a hint, |
24 |
>> Helmut. |
25 |
>> |
26 |
>> |
27 |
> |
28 |
> Another thought to. The option --with-bdeps y may help. It may not but |
29 |
> it could be worth a try. I have ran into problems where that helped and |
30 |
> have read where a few others have ran into similar problems and that |
31 |
> fixed it. It just tells emerge to dig a little deeper from my |
32 |
> understanding. |
33 |
> |
34 |
> Can't think of anything else at the moment. |
35 |
> |
36 |
|
37 |
Thanks Dale, --with-bdeps y seems to help. |
38 |
|
39 |
By the way, the '-j <n>' option of emerge has nothing to do |
40 |
with the MAKEOPTS option. |
41 |
The -j option of emerge require parallel emerging of several packages at |
42 |
a time while the -j option in MAKEOPTS requires parallel make of a |
43 |
single package. |
44 |
|
45 |
In some phases (like configure, bzip of the man pages, etc) emerge |
46 |
cannot use multiple cores for a single package. Therefore the -j option |
47 |
of emerge tries to emerge several packages at a time unless dependencies |
48 |
forbid that. It's quite clever in that respect. |
49 |
|
50 |
Helmut. |
51 |
|
52 |
-- |
53 |
Helmut Jarausch |
54 |
|
55 |
Lehrstuhl fuer Numerische Mathematik |
56 |
RWTH - Aachen University |
57 |
D 52056 Aachen, Germany |