| 1 |
On Tue, 2009-08-25 at 11:42 +0200, Fabian Groffen wrote: |
| 2 |
> Hi Stelian, |
| 3 |
> |
| 4 |
> On 25-08-2009 02:58:09 +0200, Stelian Ionescu wrote: |
| 5 |
> > Using the Solaris guide I managed to bootstrap Prefix on CentOS 5.3, |
| 6 |
> > with only a few minor problems: |
| 7 |
> > 1) I think that the guide should explicitly mention that after emerging |
| 8 |
> > binutils and gcc one should run «source $EPREFIX/etc/profile». Even if |
| 9 |
> > there's an elog message that states that, it was easy to miss it(after |
| 10 |
> > binutils) and that lead to a broken gcc |
| 11 |
> |
| 12 |
> why is that necessary? |
| 13 |
|
| 14 |
I'm not sure, but without it I managed to get a broken toolchain: when |
| 15 |
compiling coreutils with it I noticed that several of the binaries |
| 16 |
compile by the configure script(all named conftest) would loop |
| 17 |
endlessly, while compiling the relative conftest.c with the system |
| 18 |
compiler worked |
| 19 |
|
| 20 |
> > 2) I needed to export PYTHONPATH=$EPREFIX/usr/lib/portage/pym/ in order |
| 21 |
> > for the bootstrap portage to work |
| 22 |
> |
| 23 |
> can that be related to the symlink problem in the bootstrap-prefix |
| 24 |
> script? |
| 25 |
|
| 26 |
Yes, probably. With that bug fixed, the bootstrap now works fine |
| 27 |
|
| 28 |
-- |
| 29 |
Stelian Ionescu a.k.a. fe[nl]ix |
| 30 |
Quidquid latine dictum sit, altum videtur. |
| 31 |
http://common-lisp.net/project/iolib |
Archives