| 1 |
On Fri, Aug 08 2014, Alan McKinnon wrote: |
| 2 |
|
| 3 |
> This new dynamic dep stuff allows portage a degree of freedom in |
| 4 |
> selecting such providers and sometimes it decides to use the one you |
| 5 |
> don't have. To satisfy that decision, it must then uninstall what you |
| 6 |
> have. |
| 7 |
|
| 8 |
This I understand |
| 9 |
|
| 10 |
> I suspect the root cause of such apparently random behaviour is that |
| 11 |
> portage is loading your installed apps into an associative array, and |
| 12 |
> the order if items in those are random. |
| 13 |
|
| 14 |
This does surprise me. Are entries hashed with a random value? |
| 15 |
|
| 16 |
> Two solutions:: |
| 17 |
> |
| 18 |
> 1. use --backtrack <some big number> |
| 19 |
> this tells portage to search deeper and hopefully realize you |
| 20 |
> already have libav |
| 21 |
> |
| 22 |
> 2. Put libav in world, this will stop portage from trying to be helpful |
| 23 |
> |
| 24 |
> I prefer the second choice as it makes things very explicit. |
| 25 |
|
| 26 |
It looked as thought the successful choice is for portage to uninstall |
| 27 |
libav so I am not sure inserting it into world would be helpful. I |
| 28 |
can't test it since the emerge has completed successfully (at least |
| 29 |
without complaint). |
| 30 |
|
| 31 |
thanks, |
| 32 |
allan |
Archives