| 1 |
Am Samstag, 16. August 2003 12:20 Paul de Vrieze wrote: |
| 2 |
> Unfortunately dynamic dependencies in this style are not possible. |
| 3 |
|
| 4 |
Why not? Please, could you explain me what will break? |
| 5 |
(or show me the lines in ebuild.sh were things will break?) |
| 6 |
|
| 7 |
On my machine the generation of the dependencies work (Portage 2.0.48-r5 |
| 8 |
(default-x86-1.4, gcc-3.2.3, glibc-2.3.2-r1)). |
| 9 |
I tried it with my security-1.0.ebuild. |
| 10 |
I did some quick tests with several package combinations and the only thing I |
| 11 |
noticed was, that the compile and install stages cause some heavy cpu cycles, |
| 12 |
which I can not explain. (maybe the empty SRC_URI var, but why?) |
| 13 |
|
| 14 |
> I also |
| 15 |
> believe that a more fundamental solution should be implemented. Probably in |
| 16 |
> emerge itself. |
| 17 |
|
| 18 |
I agree on that for future developments. But when will that be implemented? I |
| 19 |
tried a simple solution for now. I really like to see an 'security' argument |
| 20 |
like 'system' or 'world' in Portage 2.x. |
| 21 |
OTOH I wish to keep server administration simpler now, because easier security |
| 22 |
administration is nothing, which must wait for tomorrow (imho). |
| 23 |
|
| 24 |
>It should not really be hard. It should basically do what |
| 25 |
> you propose from a dynamically generated list of insecure ebuilds. |
| 26 |
|
| 27 |
ACK. |
| 28 |
|
| 29 |
Karsten |
| 30 |
|
| 31 |
|
| 32 |
-- |
| 33 |
gentoo-dev@g.o mailing list |
Archives