| 1 |
dev-util/objprelink depends on a /usr/lib/libiberty.a that includes |
| 2 |
'htab_find_slot_with_hash' as well as other 'htab_*' things. |
| 3 |
|
| 4 |
PROBLEM 1: |
| 5 |
If this dependency is unresolved, the ebuild blindly continues, causing |
| 6 |
later qt ebuilds to fail mysteriously. The objprelink ebuild has now |
| 7 |
been fixed in dev-util/objprelink-0-r1.ebuild. |
| 8 |
|
| 9 |
But why would the libiberty dependency be unresolved, when the |
| 10 |
objprelink ebuild clearly states that it depends on sys-devel/gcc |
| 11 |
and sys-devel/binutils? |
| 12 |
|
| 13 |
PROBLEM 2: |
| 14 |
Both gcc-2.95.3-r5 and binutils-2.11.90.0.27 install (different) |
| 15 |
/usr/lib/libiberty.a's, and the one gcc installs doesn't have the |
| 16 |
'htab_*' functions that objprelink needs. |
| 17 |
|
| 18 |
This problem is NOT solved. It seems to me that gcc shouldn't be |
| 19 |
installing a broken libiberty, but what do I know? It also seems that |
| 20 |
portage could try to catch this sort of problem during 'merge'. It'd be |
| 21 |
nice, anyway. |
| 22 |
|
| 23 |
As a work-around, I re-merged binutils, which over-wrote gcc's |
| 24 |
libiberty, and now objprelink builds fine. |
| 25 |
|
| 26 |
PROBLEM 3: |
| 27 |
This isn't directly related, but I noticed along the way that portage |
| 28 |
seems to think that sys-devel/binutils-2.11.90.0.7 is newer than |
| 29 |
sys-devel/binutils-2.11.90.0.27 -- that doesn't seem right. |
| 30 |
|
| 31 |
--Chouser |
Archives