| 1 |
On 08/26/2018 12:14 AM, Zac Medico wrote: |
| 2 |
> On 08/24/2018 04:56 PM, Francesco Riosa wrote: |
| 3 |
>> |
| 4 |
>> |
| 5 |
>> Il giorno sab 25 ago 2018 alle ore 01:45 Zac Medico <zmedico@g.o |
| 6 |
>> <mailto:zmedico@g.o>> ha scritto: |
| 7 |
>> |
| 8 |
>> On 08/24/2018 04:14 PM, Francesco Riosa wrote: |
| 9 |
>> > |
| 10 |
>> > Il 24/08/18 19:08, Mike Gilbert ha scritto: |
| 11 |
>> >> On Fri, Aug 24, 2018 at 10:45 AM Kent Fredric <kentnl@g.o |
| 12 |
>> <mailto:kentnl@g.o>> wrote: |
| 13 |
>> >>> On Fri, 24 Aug 2018 10:27:01 -0400 |
| 14 |
>> >>> Mike Gilbert <floppym@g.o <mailto:floppym@g.o>> wrote: |
| 15 |
>> >>> |
| 16 |
>> >>>> If you want to define behavior that can be relied upon in |
| 17 |
>> ebuilds, it |
| 18 |
>> >>>> should be specified in PMS. PMS does not define any meaning for the |
| 19 |
>> >>>> "test" USE flag. |
| 20 |
>> >>> We should eschew idealism about how the world *should* behave, |
| 21 |
>> and avoid |
| 22 |
>> >>> making portage a steaming garbage heap in order to comply with a |
| 23 |
>> >>> terrible PMS specification of a heavily used feature. |
| 24 |
>> >> Portage still works just fine for most people who would enable |
| 25 |
>> >> FEATURES=test. Stop exaggerating. |
| 26 |
>> >> |
| 27 |
>> > People enabling FEATURES=test on most but not all packages had some |
| 28 |
>> > troubles and they need to disable them in both package.{env,use} |
| 29 |
>> > I'd like to have a flag to reinstate the previous portage |
| 30 |
>> behaviour, but |
| 31 |
>> > that's probably too late |
| 32 |
>> |
| 33 |
>> Would a FEATURES setting that implies RESTRICT="!test? ( test )" for all |
| 34 |
>> ebuilds do what you want? |
| 35 |
>> |
| 36 |
>> |
| 37 |
>> "want" is too much, |
| 38 |
>> but it would have been pleasant to have something to keep portage |
| 39 |
>> dependancies working as before. |
| 40 |
>> "testrespectuse" or "usetestrespectfeatures" come to mind or another |
| 41 |
>> command line option. |
| 42 |
> |
| 43 |
> Maybe you are not using the latest version of portage? The current |
| 44 |
> behavior with sys-apps/portage-2.3.48 seems pretty sane to me, for |
| 45 |
> example see this behavior with dev-python/setuptools-40.0.0 which has |
| 46 |
> empty RESTRICT: |
| 47 |
> |
| 48 |
> $ emerge -pvq --nodeps setuptools |
| 49 |
> [ebuild R ] dev-python/setuptools-40.0.0 USE="-test" PYTHON_TARGETS="pypy pypy3 python2_7 python3_4 python3_5 python3_6 python3_7" |
| 50 |
> $ FEATURES=test emerge -pvq --nodeps setuptools |
| 51 |
> [ebuild R ] dev-python/setuptools-40.0.0 USE="test*" PYTHON_TARGETS="pypy pypy3 python2_7 python3_4 python3_5 python3_6 python3_7" |
| 52 |
> $ FEATURES=test USE=-test emerge -pvq --nodeps setuptools |
| 53 |
> [ebuild R ] dev-python/setuptools-40.0.0 USE="-test" PYTHON_TARGETS="pypy pypy3 python2_7 python3_4 python3_5 python3_6 python3_7" |
| 54 |
|
| 55 |
Oh, I see that src_test runs even with USE=-test there, so I see why |
| 56 |
people are complaining. Anyway, an implicit RESTRICT="!test? ( test )" |
| 57 |
setting would solve that. |
| 58 |
-- |
| 59 |
Thanks, |
| 60 |
Zac |
Archives