| 1 |
begin quote |
| 2 |
On Thu, 5 Feb 2004 12:20:58 +0100 |
| 3 |
Marius Mauch <genone@g.o> wrote: |
| 4 |
|
| 5 |
> > |
| 6 |
> > Or should binaries have more metadata in them, perhaps a specific |
| 7 |
> > requirement for libraries? (sheesh, then its down to RPM again. |
| 8 |
> > that's bad. ) |
| 9 |
> |
| 10 |
|
| 11 |
|
| 12 |
|
| 13 |
> Why more metadata? Can't we just run a ldd check on the binaries, |
| 14 |
> check that the relevant libraries are present, if not do a lookup in a |
| 15 |
> (to-be-created) list of library<->package mappings and add the |
| 16 |
> relevant packages as PDEPEND? (or yell at the user to provide that |
| 17 |
> library ;) |
| 18 |
|
| 19 |
|
| 20 |
Because we don't want to download the complete package, unpack, do |
| 21 |
magic on ELF binaries and .so files, then determine "oh, this won't |
| 22 |
work" .. It would make it a veeery slow process. |
| 23 |
|
| 24 |
|
| 25 |
and library<->package binary base is a bad solution. Conflicts (we have |
| 26 |
them) and other nagging things like USE flags have a role here |
| 27 |
|
| 28 |
Remember, some library dependencies are due to USE flag settings. we |
| 29 |
can't accomodate for all possible combinations.. Its not feasible. |
| 30 |
|
| 31 |
//Spider |
| 32 |
|
| 33 |
-- |
| 34 |
begin .signature |
| 35 |
This is a .signature virus! Please copy me into your .signature! |
| 36 |
See Microsoft KB Article Q265230 for more information. |
| 37 |
end |
Archives