Circuit network requests switching too fast in auto-mall / question about new circuit mechanics

[MrKeram]

I am experiencing an issue with my circuit-network-based automated mall (auto-mall). Signal requests coming from the circuit network seem to toggle/switch extremely fast, causing the system to unstable state or not function as expected.
I am not sure if this is a bug or an intended behavior change / new mechanic introduced in version 2.1 (e.g., tick updates, reader priorities, or new selector combinator mechanics). That automall work on version 2.0.

How to reproduce:

• Build an automated mall setup driven by circuit network signals (set up requesting signals via circuit).

• Connect the circuit network feed to the requestors/assemblers.

• Observe how fast the requested signals cycle or flip states under heavy or changing network demand.

What should happen:
Signal requests should update predictably without rapid toggling/flickering on short tick intervals, allowing machines/inserters to process requests smoothly.
What happens:
Requests from the circuit network switch too fast, interrupting the automation sequence or making the auto-mall switch states faster than entities can react.

Screen Recording 2026-10-04 at 01.11.10.mp4

(92.88 MiB) Downloaded 6 times

[Alfonse215]

That automall work on version 2.0.

I can't see the source of the signals, so I don't know where it's coming from. But the biggest change for 2.1 that would specifically affect automalls is the ability of devices to pick which color of wire they send outputs to or receive inputs from. This is mostly a good change, but the downside is that they simultaneously removed some code that allowed a device to send and receive through a bit of a hack called "magic subtraction".

This is particularly important for automalls, as you often want to set the recipe *and* read the inputs of said recipe. This would work in 2.0 because the assembler would recognize that it's outputting certain signals on a wire and thus "subtract" those values when looking for inputs. So if you set, for example, "advanced circuits", the assembler would output, for example, "copper cables" on the same wire that you sent the "advanced circuit" signal. The assembler would not get confused because it subtracts the signals that it is sending before reading the inputs.

Because 2.1 allows you to explicitly say which wire is input and which is output... that subtraction of outputs from inputs no longer needs to happen. So it doesn't. And if you haven't adjusted your automall setup accordingly, it can very well confuse the ingredient signals it is outputting for the recipe signal it's supposed to read.

You need to go into your automall setup and configure the wire outputs accordingly.

[MrKeram]

Thank you for the detailed explanation! That makes complete sense. I didn't know about this changes.

Thanks again for pointing me in the right direction!