| 1 |
On 8/22/06, Adrian Frith <adrian@××××××××.za> wrote: |
| 2 |
> If I am right, doesn't it mean that the whole "emerge -e system" step is |
| 3 |
> pointless? Couldn't one just do "emerge -u gcc" and then "emerge -e |
| 4 |
> world"? Or am I being stupid? |
| 5 |
|
| 6 |
You are talking about when upgrading to new versions of gcc (based on |
| 7 |
your use of the -u flag), right? If so, then you are essentially |
| 8 |
correct. |
| 9 |
|
| 10 |
Indeed what I did for the last upgrade was a slightly modified version |
| 11 |
of the gcc upgrade guide [1]: |
| 12 |
|
| 13 |
emerge -u gcc |
| 14 |
gcc-config blah-blah |
| 15 |
source /etc/profile |
| 16 |
emerge --oneshot libtool |
| 17 |
revdep-rebuild --library=libstdc++.so.6 |
| 18 |
emerge --prune gcc |
| 19 |
emerge -e world |
| 20 |
|
| 21 |
The revdep-rebuild step was to rebuild all C++ applications so they |
| 22 |
link with the new C++ library. After that, I could expect my system |
| 23 |
to be sane and survive the prune of old gcc versions. The emerge -e |
| 24 |
world was to take advantage of whatever new optimizations were |
| 25 |
available. |
| 26 |
|
| 27 |
However, for someone who doesn't really know what they are doing when |
| 28 |
upgrading gcc, there is some extra safety possible with the "emerge -e |
| 29 |
system" step, since once that completes, you *know* your critical |
| 30 |
system packages have been rebuilt and still work. If you skip that |
| 31 |
and go straight to the emerge -e world, then if something fails to |
| 32 |
build, you might end up with some critical packages left behind. |
| 33 |
|
| 34 |
-Richard |
| 35 |
|
| 36 |
[1] http://www.gentoo.org/doc/en/gcc-upgrading.xml |
| 37 |
-- |
| 38 |
gentoo-user@g.o mailing list |
Archives