1 |
On 19-04-2008 15:31:35 +0200, Bacchella Fabrice wrote: |
2 |
>> So the alternative is to switch from arch to ~arch in the post-sync |
3 |
>> stage. Thoughts here? |
4 |
>> |
5 |
>> Questioni is now how are we going to implement this, if we all agree |
6 |
>> this is the way to go? Do we agree in the first place? |
7 |
> |
8 |
> I think that's a good idea. In the last 2/3 weeks, I had too many problems |
9 |
> with package upgrading while I was building. |
10 |
> |
11 |
> I don't know if having to upgrade all system package is a big deal. For |
12 |
> example, there is a lot of /usr/sfw/lib in RUNPATH left over for the |
13 |
> solaris build. So a re-emerge is a good thing anyway. And there a lot of |
14 |
> package to get after the boot strap, so it's going to be a long process |
15 |
> anyway. |
16 |
|
17 |
Method A: |
18 |
emerge a b c |
19 |
emerge --sync |
20 |
emerge -u system |
21 |
emerge -e system |
22 |
echo 'ACCEPT_KEYWORDS="~arch"' >> make.conf |
23 |
emerge -e system |
24 |
|
25 |
Method B: |
26 |
emerge a b c |
27 |
emerge --sync |
28 |
emerge -u system |
29 |
echo 'ACCEPT_KEYWORDS="~arch"' >> make.conf |
30 |
emerge -e system |
31 |
|
32 |
|
33 |
Method A has the clear advantage that you'll much more likely make it to |
34 |
the end of the bootstrap document, however once you decide you want to |
35 |
do more than just system packages (quite likely), you'll basically |
36 |
confronted with the entire system set having newer versions (gcc, |
37 |
binutils, python, coreutils, to name a few larger ones). |
38 |
|
39 |
Method B is more risky, but should work if we manage not to break the |
40 |
life tree, which in theory shouldn't happen, but in practise not always |
41 |
is. Though should be easy to recover from that using emerge --sync. |
42 |
|
43 |
|
44 |
-- |
45 |
Fabian Groffen |
46 |
Gentoo on a different level |
47 |
-- |
48 |
gentoo-alt@l.g.o mailing list |