my opinions

[antkss]

why don't run any ai request in threads, it will be more efficiency

[mahaloz]

PRs for it are welcome :).

The reason they were not originally there is for this reason:

• When you run a prompt on a decompiled function you are on main thread. That is good news, because it all modern decompilers you need to be on main thread to make a change to decompilation (like re-typing a variable).
• If you thread, now you need to wait in another thread until the AI request is done, then modify the users decompilation. This will, depending on the size of the function, randomly freeze the users screen to make the change when it happens (recall, you need to be on main thread to edit decompilation)
• It also runs the problem of a race condition: what if the user modifies their own function before the AI finishes. Now changes get overwritten, or, in the worst case, the decompiler crashes because of a UAF on the data (happened in IDA Pro before).