-
Notifications
You must be signed in to change notification settings - Fork 365
Open
Description
When window.openai.sendFollowUpMessage is called before the LLM has finished responding to the user, the call is simply ignored without any error.
There is also no way to determine whether the LLM has finished responding or not.
Because of this, we have a problem: if the user taps a button in the widget before the LLM has finished its response, we have no way to handle that action.
I found only one workaround to detect whether sendFollowUpMessage was handled or ignored — measuring the function’s execution time. If it returns in under ~50 ms, it was ignored; if it takes longer, it was handled. But this is a very poor solution.
Could you either return an error when the call is ignored, or queue the call until the LLM finishes responding?
xarsh
Metadata
Metadata
Assignees
Labels
No labels