I think the solution to this was already addressed and PR'ed by megagrump which is now in the `master` and `v1.0.0` branches
Nope. AFAIK, there is no change in this matter in both branches for now... I want to be wrong, however.
Feel free to make PR It would be easier if we can compare the performance of the current `master` branch and your changes
It's not matter of performance but responsiveness and reliability. And it would mean deep architectural cganfes, as Slab by its design rely on updated state, and events needs some kind of callbacks or message queue. It's something completely different from immediate mode GUI Slab is. So, it would be hard...
From the code patch, it appears that events were dropped because the table was iterated in forwards order, not in backwards order. It also appears that the events were treated as a stack, not a queue.
I have finally found the code and way Slab handles events. It's by overriding handlers directly. To me, it's very bad idea, as I may wish to handle events as well. But Slab takes everything for itself. I strongly recommend to enable forwarding funtions instead of overriding handlers.
I have finally found the code and way Slab handles events. It's by overriding handlers directly. To me, it's very bad idea, as I may wish to handle events as well. But Slab takes everything for itself. I strongly recommend to enable forwarding funtions instead of overriding handlers.
pirogronian wrote: ↑Sat Oct 29, 2022 9:33 am
I have finally found the code and way Slab handles events. It's by overriding handlers directly. To me, it's very bad idea, as I may wish to handle events as well. But Slab takes everything for itself. I strongly recommend to enable forwarding funtions instead of overriding handlers.
PRs are welcome for the proposed changes
Ok, my another realisation: Slab actually calls all previously defined handlers by itself. It works with statically defined callbacks, but I'm unsure about other side effects of such a behaviour. Maybe it should be mentioned in documentation to avoid accidental and surprising misbehaviour if user will manipulate with callbacks more dynamically.
It would be easier if we can compare the performance of the current `master` branch and your changes