| 1 |
On 01/26/2015 03:57 AM, Alexis Ballier wrote: |
| 2 |
> On Fri, 23 Jan 2015 09:19:19 -0800 |
| 3 |
> Zac Medico <zmedico@g.o> wrote: |
| 4 |
> |
| 5 |
>>> what Michał claims is that 'foo' will want ffmpeg, no matter what. |
| 6 |
>> |
| 7 |
>> Maybe he's talking about some case where portage failed to trigger a |
| 8 |
>> rebuild when appropriate. We've had a number of bugs like this that |
| 9 |
>> have been fixed. The most recent recent example is bug 531656 [1]. |
| 10 |
> |
| 11 |
> Yes I know about this one :) To be honest I find it much more |
| 12 |
> difficult and convoluted (?) than the present case, yet you fixed it. |
| 13 |
|
| 14 |
I like to handle as many cases as practically possible, because the same |
| 15 |
heuristics may be applicable to many other cases that haven't been |
| 16 |
reported yet. |
| 17 |
|
| 18 |
> Even worse: there has been tree-wide changes made because of this |
| 19 |
> (removal of := from all || ( ffmpeg:= libav:= )) instead of filling |
| 20 |
> portage bugs... which I consider just plain wrong and was in fact my |
| 21 |
> initial rant. |
| 22 |
|
| 23 |
Well, the ability of portage to handle a particular type of dependency |
| 24 |
construct does not necessarily imply that this type of dependency |
| 25 |
construct should be encouraged. |
| 26 |
-- |
| 27 |
Thanks, |
| 28 |
Zac |
Archives