> 80 + callbacks->self = nullptr;
>
> Doesn't 'self' need be atomic?
yes, it should be, fixed.
> 45 + hwc_wrapper = std::make_shared<mga::RealHwcWrapper>(hwc_native,
> logger);
>
> Perhaps have a mga::ResourceFactory::create_hwc_wrapper() instead (and then we
> probably won't need ResourceFactory::create_hwc_native_device()).
Yes, this is the direction I want to go, to hide the hwc type behind the wrapper. There are a few interfaces that still need to be plumbed out on HwcWrapper before getting rid of the create_hwc_native_device() function, namely detecting the version, and querying some of the other hwc properties. I considered that a bit beyond the scope of this bug fix, so will leave it for future improvement.
> 80 + callbacks->self = nullptr;
>
> Doesn't 'self' need be atomic?
yes, it should be, fixed.
> 45 + hwc_wrapper = std::make_ shared< mga::RealHwcWra pper>(hwc_ native, ctory:: create_ hwc_wrapper( ) instead (and then we ::create_ hwc_native_ device( )).
> logger);
>
> Perhaps have a mga::ResourceFa
> probably won't need ResourceFactory
Yes, this is the direction I want to go, to hide the hwc type behind the wrapper. There are a few interfaces that still need to be plumbed out on HwcWrapper before getting rid of the create_ hwc_native_ device( ) function, namely detecting the version, and querying some of the other hwc properties. I considered that a bit beyond the scope of this bug fix, so will leave it for future improvement.