| 1 |
On 23-03-2006 04:14:17 +0000, m h wrote: |
| 2 |
> On 3/23/06, Kito <kito@g.o> wrote: |
| 3 |
> > > To get around the gcc issue, gcc3.3.6 was installed. |
| 4 |
> > |
| 5 |
> > Ok, so we'll bump the gcc versions in the prefix tree. I'd like to |
| 6 |
> > 4.x in there anyway as well. |
| 7 |
> |
| 8 |
> This was NOT a version bump. The version in the tree was 3.4.4... |
| 9 |
|
| 10 |
... which works fine on amd64 here... |
| 11 |
|
| 12 |
(Ok, I screwed it up, but that's my own fault.) |
| 13 |
|
| 14 |
I do have a question though. |
| 15 |
|
| 16 |
Did you install binutils, gcc, linux-headers and glibc from portage now? |
| 17 |
The current stuff appears to work, but I think it has some paths set |
| 18 |
wrong, therefore using still headers from / instead of ${PREFIX}. |
| 19 |
|
| 20 |
I'm on a fix here, if I can get my gcc recompiled (I screwed up the |
| 21 |
linker: ${PREFIX}/usr/bin/ld: crt1.o: No such file: No such file or directory) |
| 22 |
|
| 23 |
> > > The next issue was complaints about "-fno-stack-protector" available |
| 24 |
> > > on gcc. The was no gcc binary in $PREFIX, so I resolved this by |
| 25 |
> > > adding a gcc symlink in $PREFIX to ${PREFIX}/usr/bin/gcc-3.3.6. |
| 26 |
> > |
| 27 |
> > This should be solved by using gcc-config, which sets the appropriate |
| 28 |
> > symlinks. Its probably broken in prefix though, haven't tested it yet. |
| 29 |
|
| 30 |
Yeah, but this tool doesn't work (yet). We need it, because maintaining |
| 31 |
the symlinks yourself is a tedious job. |
| 32 |
|
| 33 |
> > > I hit the missing libpython2.4.so issue again, so I added |
| 34 |
> > > ${PREFIX}/usr/lib/ to LD_LIBRARY_PATH. |
| 35 |
> > |
| 36 |
> > Hmm, so Diego was right...this obviously needs to be fixed. did |
| 37 |
> > running env-update work as expected? |
| 38 |
|
| 39 |
It probably runs fine, but as long as you don't use bash, tcsh or zsh |
| 40 |
from the prefix, it won't source it. You need to emerge one of those |
| 41 |
shells (at least for tcsh I am 100% sure ;) ) and execute it, because |
| 42 |
they are properly patched/configured to use the prefix to look for |
| 43 |
system-wide init files. Only with such shell you get the contents of |
| 44 |
"env-update", a.k.a. env.d/* |
| 45 |
|
| 46 |
> > > That along with a few ebuild foo.ebuild digest, and manually |
| 47 |
> > > downloading a few tarballs and putting them in distfiles allowed me to |
| 48 |
> > > finish emerge -av system. |
| 49 |
> > |
| 50 |
> > Why the manual downloads? |
| 51 |
> |
| 52 |
> Don't know but some of the tarballs weren't found in the default |
| 53 |
> places. (I don't think it was an exhaustive search, but after I see 5 |
| 54 |
> or so failed attempts I just google for "Index of" and the tarball |
| 55 |
> name, and download it and put it in distfiles. |
| 56 |
|
| 57 |
Hmmm, ok, maybe the mirror select thinghy would come in handy here. I |
| 58 |
noticed this myself too, that it quite often hangs on slow servers or |
| 59 |
just can't find the file on many mirrors it tries. |
| 60 |
|
| 61 |
> > > Next task, get apache working. Stay tuned.... |
| 62 |
|
| 63 |
wow! That's freaky! (I'm just working on openssh -> but that triggered |
| 64 |
a bug in the current gcc/binutils combination) |
| 65 |
|
| 66 |
Anyway, I'm interested in how easy you can make it work! |
| 67 |
|
| 68 |
|
| 69 |
-- |
| 70 |
Fabian Groffen |
| 71 |
Gentoo for Mac OS X Project |
| 72 |
-- |
| 73 |
gentoo-osx@g.o mailing list |
Archives