[Feature Request] Add `multishell.stop()` or a way for programs to control / detect the "Press any key to continue" prompt

[HKXingluo]

📄 Description /

Currently, when a program is launched via bg (or shell.openTab), it will display "Press any key to continue" and wait for input upon exit if the tab was never focused (bInteracted == false). This behavior is hardcoded in multishell and cannot be overridden by the program itself.

This is problematic for truly background or fire-and-forget scripts that:

• Produce no interactive output,
• Are meant to run and close automatically,
• Or are controlled entirely via other channels (e.g., rednet, APIs).

Suggested Solutions

1. Add a multishell.stop(id) function that terminates a process without triggering the "Press any key" prompt.

2. OR: Expose a flag or function like multishell.setNoWaitOnExit(id, true) to suppress the prompt.

This would give developers fine-grained control over program lifecycle and improve the usability of background tasks in multishell environments.

[Wojbie]

There is indirect way to trigger bInteracted flag using already existing methods:

local focusTabId = multishell.getFocus()
local thisTabId = multishell.getCurrent()
multishell.setFocus(thisTabId)
multishell.setFocus(focusTabId)

While it is not one line solution you are proposing, it might be helpful for other players who use current and earlier CC:T versions.

[HKXingluo]

Although this is a solution, it is not perfect/incomplete.

Two
"multishell.setFocus" calls:

For humans, there is hardly any delay, but for a program, this is a genuine switch to another window. Any keystrokes/click events during this period will be passed to the other window.

Moreover, for extremely short programs (e.g., only one line used to call another program outside the main program for a very brief moment), the code will finish running instantly, and by the time the switch occurs, it is already too late.

[Wojbie]

Any keystrokes/click events during this period will be passed to the other window.

Please correct me if i am wrong (I might be overlooking a yield somewhere in multishell code) but believe no code called during "multishell.setFocus" call would be yielding. Which due to single threaded nature of lua would mean that no other code could be running at same time, so 2 set focus switches should occur uninterrupted, without yields, so it means without having any effect on event queue or without any other code getting executed during their execution. You can see this in action using this simple test code:

local thisTabId = multishell.getCurrent()
local newID = shell.openTab("worm")
parallel.waitForAll(function()
 print("preFocus")
 multishell.setFocus(newID)
 multishell.setFocus(thisTabId)
 print("postFocus")
end,
function()
 print("other code")
end)

"other code" had no chance to execute during the switch back and forth, so it could not have received any events during incorrect focus state.

Moreover, for extremely short programs (e.g., only one line used to call another program outside the main program for a very brief moment), the code will finish running instantly, and by the time the switch occurs, it is already too late.

In this case, neither my setFocus trick, or any of your proposed solutions is usable externally to the extremely short program, because for short code that is executed, runs without a yield and finishes, no other code has opportunity to be executed. Short program itself would have to be one do setFocus trick, or call one of your methods before it ends, to ensure it don't get hit with "Press any key to continue" prompt.

[HKXingluo]

The short code I mentioned refers to code that calls functions like
"multishell.stop" within itself to terminate itself.

[HKXingluo]

Or perhaps we need a way to actively terminate other multishell programs and their windows, or method to hide the tab bar, or to make tabs hidden (not displayed in the tab bar) when opening them via "openTab".

I want to implement these features because bg truly support multithreading. That's why I'm considering whether it's possible to rewrite a custom multishell implementation in CC to achieve these goals.

[SquidDev]

This has come up a couple of times (#1909, #2083), and I feel in quite a few cases this is a bit of an A/B problem, where people are using multishell just as a way to run code in parallel (and would be better served by coroutines). What exactly are you trying to do here?