Overall, this looks like a nice simplification but there's a lot to it and (due to the poor test coverage of the 3rd_party input code) I don't feel confident that nothing breaks as a result.
It isn't easy to get the code under test (especially without reworking it - which is what this MP does), so I can't insist on "more tests".
Overall, this looks like a nice simplification but there's a lot to it and (due to the poor test coverage of the 3rd_party input code) I don't feel confident that nothing breaks as a result.
It isn't easy to get the code under test (especially without reworking it - which is what this MP does), so I can't insist on "more tests".
Anyway, there's a bit of tidy up needed...
226 + synthesizeCance lationEventsFor InputChannelLoc ked( nnel, options);
227 + focusedInputCha
Strange indentation
~~~~
2845 + virtual ~InputRegistrar() noexcept(true) {}
2978 + virtual ~InputTargetEnu merator( ) noexcept(true) {}
These don't gain by being inline
~~~~
2853 +protected: const InputRegistrar&) = delete;
2854 + InputRegistrar(
2855 + InputRegistrar& operator=(const InputRegistrar&) = delete;
2982 +protected: erator( const InputTargetEnum erator& ) = delete; erator& operator=(const InputTargetEnum erator& ) = delete;
2983 + InputTargetEnum
2984 + InputTargetEnum
No need for a protected section for these - and aren't they deleted in the base anyway?
~~~~~
2858 + droidinput: :sp<droidinput: :InputDispatche rInterface> input_dispatcher;
should be const