1 |
On 8/21/06, Stefan G. Weichinger <lists@×××××.at> wrote: |
2 |
> Richard Fish schrieb: |
3 |
> > or "emerge -e world" to complete. I would be tempted to just change |
4 |
> > the flags and hold off on recompiling everything until the next |
5 |
> > version of gcc comes out. |
6 |
> |
7 |
> ( ... "next version" in terms of minor- or major-version?) |
8 |
|
9 |
Gcc version numbers are in the form of X.Y.Z, so at a change in X or |
10 |
Y. The .Z changes should be just bug-fixes. The sad thing is that, |
11 |
due to changes in the C++ library, it is commonly to recompile all C++ |
12 |
applications when a new version of gcc comes out. And even if it |
13 |
isn't strictly /necessary/, an emerge -e world is considered the safe |
14 |
way to handle a gcc upgrade. |
15 |
|
16 |
> I see the point in this. (AFAIK there is no way to break up "emerge -e |
17 |
> xy" into smaller pieces, something to do in several separated steps. |
18 |
|
19 |
Actually there is. You can find all packages not compiled with -Os |
20 |
and rebuild them with something like: |
21 |
|
22 |
cd /var/db/pkg |
23 |
for pkg in */* ; do |
24 |
grep -v -- "-Os" $pkg/CFLAGS >/dev/null |
25 |
test $? -eq 0 && emerge --oneshot =$pkg |
26 |
done |
27 |
|
28 |
> From your posting I conclude that it also won't do any harm to re-emerge |
29 |
> selected parts with new CFLAGS?) |
30 |
|
31 |
Correct. |
32 |
|
33 |
> Apart from this I have enough computer-related experience to know that I |
34 |
> simply should be happy with the |
35 |
> luks-encrypted/cpufreq'ed/hibernating/etc. gentoo-system I now have at |
36 |
> hand, instead of spending numerous hours to gain minimal speedups. |
37 |
|
38 |
Hrm, I also have the experience, but apparently not the good sense.... :-P |
39 |
|
40 |
-Richard |
41 |
-- |
42 |
gentoo-user@g.o mailing list |