On Tue, Nov 17, 2020 at 12:00 PM Martin Jansa <[email protected]> wrote:
>
> And there are some cases where we need only native recipe and creating a 
> target one just to BBCLASSEXTEND seems worse than the current state with 
> native.bbclass.
>
> Or there are some cases where the native build is so significantly different 
> from the target one, that using BBCLASSEXTEND just makes it more difficult to 
> read without any benefit.
>
> So please don't remove the option to just inherit native if there aren't any 
> technical reasons to do that.
>

I think we have a patch to address the fallout so we should be good here now.

> On Tue, Nov 17, 2020 at 7:09 PM Khem Raj <[email protected]> wrote:
>>
>> On Tue, Nov 17, 2020 at 9:33 AM Alexander Kanavin
>> <[email protected]> wrote:
>> >
>> > On Tue, 17 Nov 2020 at 16:41, Khem Raj <[email protected]> wrote:
>> >>
>> >> On Tue, Nov 17, 2020 at 1:19 AM Alexander Kanavin
>> >> <[email protected]> wrote:
>> >> >
>> >> > Yes - and if python3-scons-native is the only such recipe we can find, 
>> >> > I think it's better to merge the recipe with the target one.
>> >> >
>> >> python3-scons-native is just a manifestation of a larger problem which
>> >> is a regression and this suggestion compounds the problem by setting
>> >> an unequal rule where non-python packages can
>> >> have a native-recipe and recipe but python ones can not, therefore we
>> >> need to fix this properly in core classes instead.
>> >
>> >
>> > If non-python native-only recipes inherit something that DEPENDS on a 
>> > target item, they will have the exact same problem. This is what needs 
>> > fixing - but not by special-casing DEPENDS in every class where it 
>> > happens, or might happen.
>> >
>>
>> keep this issue aside and think in a generic way about how recipes can
>> be written regardless of native or not. behavior has to be consistent
>>
>> > I'd rather convert all native-only recipes to target ones + BBCLASSEXTEND, 
>> > so it works the same way everywhere, and minimizes surprises.
>> >
>>
>> that's wishful, there are a lot of layers that will have to also
>> change, and we have to change the semantics of native recipes where
>> BBCLASSEXTEND is the only way to do them.
>>
>>
>> > Alex
>> >
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#144753): 
https://lists.openembedded.org/g/openembedded-core/message/144753
Mute This Topic: https://lists.openembedded.org/mt/78311501/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to