| 1 |
On 13/03/12 02:23, Neil Bothwick wrote: |
| 2 |
> On Tue, 13 Mar 2012 01:54:30 +0200, Nikos Chantziaras wrote: |
| 3 |
> |
| 4 |
>>>> Anyone care to offer an opinion on what it will take to get PROVIDES |
| 5 |
>>>> support in portage? |
| 6 |
>>> |
| 7 |
>>> IMO, it would take virtuals causing so many headachy breakages that |
| 8 |
>>> some devs started keeping up a steady drumbeat on irc and mailing |
| 9 |
>>> lists. When the number of virtual packages gets close to a thousand, |
| 10 |
>>> I'd guess that might happen. Then there would be years of discussion |
| 11 |
>>> and GLEP proposals, and by EAPI 207 it should be done. |
| 12 |
>> |
| 13 |
>> The problem isn't the amount of virtuals. This doesn't affect the |
| 14 |
>> users much. It's the inability for people to offer replacement |
| 15 |
>> packages in overlays. |
| 16 |
> |
| 17 |
> They could include a modified virtual in the overlay, but your point is |
| 18 |
> valid; including the information in the ebuilds is more flexible. |
| 19 |
|
| 20 |
This only works if portage has a virtual. If it doesn't, you're |
| 21 |
screwed. You need to also provide modified packages of all ebuilds |
| 22 |
depending on the package you're offering a replacement for. As you can |
| 23 |
guess, it's not practical. |
| 24 |
|
| 25 |
This leaves only one option; have users put the original package in |
| 26 |
package.provided and emerge your replacement as a non-dep (going in |
| 27 |
"world".) |
Archives