At last year's FOSDEM, we discussed the feasibility of Wayland input methods. One year later and we have a working solution that will be part of Wayland's next stable release. What better place than FOSDEM (again!) to show the progress and discuss the remaining work?
There is a reason why XIM -- X11's input method -- is still popular, 18 years after its initial design: A unified API that works across toolkits is highly appealing to developers of CJK input methods. Even today, input method frameworks are expected to have decent XIM support.
The downsides of XIM and X11 are the complexity and the lack of protocol evolution over the last years. Everyone who tried to integrate input methods with an X11 window manager will know the pain of focus stealing and window stacking issues. Initiatives to improve X11 in that area have stalled for years. In the meantime, toolkits have come up with their own solutions, ignoring the fragmentation caused for input method developers.
On the other hand, Android's input method framework is a good showcase for modern use-cases (think virtual keyboard) and the tight integration with the system compositor.
With Wayland, we can achieve the same.
The talk will be an open invitation to device manufacturers, toolkit providers and input method developers alike to give Wayland and its new input method protocol a try and help us with feedback.
