| 1 |
Mikael Hallendal wrote: |
| 2 |
>tis 2002-01-08 klockan 16.54 skrev Damon M. Conway: |
| 3 |
> |
| 4 |
>> I think this is where eclasses could really help. Eclasses should allow |
| 5 |
>> you to create a meta ebuild that looks for certain make.conf vars set and |
| 6 |
>> react accordingly. danarmak and drobbins are the ones to ask for more |
| 7 |
>> details on eclasses. |
| 8 |
> |
| 9 |
>For now eclasses should _only_ be used in KDE. A decission still has to |
| 10 |
>be made about eclasses. And both me and drobbins have some objections to |
| 11 |
>eclasses. |
| 12 |
|
| 13 |
Hmm, ok. I thought they were on their way to finalization. |
| 14 |
|
| 15 |
>For one I think that eclasses break one of ebuilds most important |
| 16 |
>strength. The ease of use, that they are almost identical to installing |
| 17 |
>a package manually. |
| 18 |
|
| 19 |
Depends on usage. I think eclasses could be quite powerful for |
| 20 |
cross-platform capabilities and system types (SuxOS, Web Server, etc), and |
| 21 |
they, in theory, can pick up the global configuation baton where make.conf |
| 22 |
leaves off. As long as they are well-defined, I don't see how they really |
| 23 |
cause any problems, or make ebuilds any harder to write. |
| 24 |
|
| 25 |
kabau |
| 26 |
|
| 27 |
-- |
| 28 |
"UNIX was not designed to stop you from doing stupid things, because that |
| 29 |
would also stop you from doing clever things." --Doug Gwyn |
Archives