[1.1.34] circuit network belt reading wrong/missed on merging segment

[bendog]

Apologize first if this is not a bug but intended behavior, but I found it very un-intuitive at least and couldn't find an explanation on the wiki.

what happened

screenshot-tick-53258.png (7.56 MiB) Viewed 824 times

So basically I'm trying to count items on the main bus by tracking all the input/output branches. Here you can see two minimal examples (A and B), where the connected belt segments send out contents in pulse mode and a simple memory cell is used to track items that went into the bus. The only difference between the two examples is the location of the reading belt segment.

Now, the bottom example (B) works exactly as expected, where the signal in the memory cell matches the item that goes through the belt, as can be verified by the content in the chest. However, there is a mismatch when using the top example (A). Specifically, when there are items on the top lane of A2, and they merge/mix with those coming from A1 at the reading segment, some items from A1 will be missed during the counting. In the attached save file, I have placed steel plates on A2/B2 and copper plates on A1/B1, and you can see the count of copper plates is 1 count smaller than those in the chest for bus A, but not bus B. In my real game save this error can accumulate over time.

what I expect

Both A and B bus should work as expected and produce signals corresponding to the content of chests.

to reproduce

1. clear chest content and memory cell (you can turn on the constant combinator to clear memory)
2. put lots of items on the left end of the top input belt (A1 or B1)
3. while items on the top input belt is still flowing into the main bus, put lots of items (preferrably another type) on the top lane at the left end of the main bus/bottom belt (A2 or B2). The items from the two belt has to be merging/mixing on the same lane to reproduce this bug
4. check signal of the memory cell and the content of the chest, you should see mismatch for A but not for B

Let me know if this makes sense or if anyone can reproduce this!

latest log:

factorio-current.log

(6.19 KiB) Downloaded 78 times

game save:

belt_bug.zip

(2.32 MiB) Downloaded 76 times

[Loewchen]

viewtopic.php?f=23&t=84428

[bendog]

oh wow that sounds like exactly what's happening. Thanks for the quick reply!

I guess personally I still think this is a surprising behavior to ppl that couldn't find that particular thread (like me...). When I read "items inserted or taken by external entities" I was picturing inserters messing with the belt and obviously that's a bad idea. But somehow another belt coming in feels like "part of the system" and sounds like it could work (also it does work, just not when saturated). So maybe a bit more explicit explanation of this either in the tooltip or wiki can save someone else a few hours of frustration (and duplicated bug report)?

Anyway, feel free to mark this as closed. Thanks again for the reply!