| 1 |
As a result of Cardoes earlier mail we talked a bit about possible |
| 2 |
solutions in #gento-portage, and I suggested to let portage |
| 3 |
automatically inject the deps based on SRC_URI pattern matching. |
| 4 |
A mapping of extensions and their unpack deps would be kept in the tree |
| 5 |
(e.g. mapping '.tar.bz2' to '( app-arch/tar app-arch/bzip2 )' |
| 6 |
|
| 7 |
Benefits: |
| 8 |
- simplified depstrings for most packages (I assume >90% of the tree) |
| 9 |
- easier maintenance if dependencies are changing, or additional |
| 10 |
implementations become supported (this could also be achieved with |
| 11 |
virtuals though) |
| 12 |
- potentially more accurate dependencies, as some common errors would |
| 13 |
be eliminated (e.g. proper treatment of use-conditionals, not adding |
| 14 |
unpack deps to RDEPEND) |
| 15 |
- long-term adds the possibility to remove bzip2, gzip and tar from |
| 16 |
@system |
| 17 |
|
| 18 |
Potential problems: |
| 19 |
- might cause trouble for some packages that use custom code for |
| 20 |
unpacking, or due to circular deps, this could simply be solved with a |
| 21 |
new RESTRICT value though. |
| 22 |
- automagic deps could be confusing to devs/users |
| 23 |
- not available for existing EAPIs (due to the mentioned problems) |
| 24 |
- slightly slower dep calculations due to the extra processing |
| 25 |
- possible match errors |
| 26 |
|
| 27 |
So, is this something ebuild maintainers would like in general, or does |
| 28 |
such a feature cause you nightmares? |
| 29 |
|
| 30 |
Marius |
| 31 |
-- |
| 32 |
gentoo-dev@l.g.o mailing list |
Archives