1 |
On Tue, 2006-08-22 at 20:36 +0200, Adrian Frith wrote: |
2 |
> On Tue, 2006-08-22 at 16:12 +0200, Alan Mckinnon wrote: |
3 |
> > modify CFLAGS in /etc/make.conf |
4 |
> > emerge -e system |
5 |
> > emerge -s world |
6 |
> > |
7 |
> > This will rebuild your toolchain (gcc, glibc and friends) to use -O2 |
8 |
> > then rebuild the entire system, including the toolchain again, with -O2. |
9 |
> > Your current compiler was built with -O3, and you want to rebuild the |
10 |
> > system using a compiler compiled as -O2, hence the 2 step process. |
11 |
> |
12 |
> Correct me if I am wrong, but doesn't the gcc ebuild in fact use the |
13 |
> "make bootstrap" method of compiling gcc? - that is: |
14 |
> 1. compile with installed compiler -> stage1 |
15 |
> 2. compile with stage1 compiler -> stage2 |
16 |
> 3. compile with stage2 compiler -> stage3 |
17 |
> 4. compare stage2 and stage3 and install stage3 if they are the same |
18 |
> |
19 |
> If I am right, doesn't it mean that the whole "emerge -e system" step is |
20 |
> pointless? Couldn't one just do "emerge -u gcc" and then "emerge -e |
21 |
> world"? Or am I being stupid? |
22 |
|
23 |
I believe the gcc ebuild does do a bootstrap compile, which does imply |
24 |
that emerge -e system is redundant. |
25 |
|
26 |
But, we had a huge long rambling thread on this point recently which I'd |
27 |
rather not rehash again :-) and IIRC the general consensus was that |
28 |
'emerge -e system ; emerge -e world' was the way to go, for reasons |
29 |
broader than just gcc. I forget the details, maybe we should both check |
30 |
the archives. |
31 |
|
32 |
What I do know is that 'emerge -e system ' is relatively quick compared |
33 |
to 'emerge -e world', so the extra step didn't hurt much when I recently |
34 |
upgraded gcc, it was something like 2 hours for the one, and 36 for the |
35 |
other (!) |
36 |
|
37 |
alan |
38 |
|
39 |
-- |
40 |
gentoo-user@g.o mailing list |