[2.1.9] Filtered splitters don't filter properly

[SoYman]

I have a build that separates the products of recycled processing units of all different quality levels.
The splitter should filter the electronic circuits to the left and the advanced circuits to the right. Here are two screenshots from the build:

Screenshot_20260710_141002.png (226.26 KiB) Viewed 379 times

Screenshot_20260710_141017.png (198.17 KiB) Viewed 379 times

As you can see, in the first screenshots there are electronic circuits that weren't filtered correctly and landed on the right. Similarly, on the second screenshot, an advanced circuit passed the filter and ended up on the left side. There are many other instances of this 4x4 buffer setup and this issue seems to occur sporadically on most of them.

My expectation is that green circuits are filtered to the left and red to the right. This occurs rarely, watching the build running shows no example of this, but when the build is left to run for dozens of minutes, some of the 4x4 buffers will have had this issue, as shown.

Here is the save this happens in https://drive.google.com/file/d/1yxU6Vy ... sp=sharing Sorry, it's big.

Here is the location of the build in the save, circled in cyan.

Screenshot_20260710_145928.png (1.68 MiB) Viewed 379 times

[boskid]

This is suspicious because your build also has additional splitters near the recycler output which should sort greens on one side and reds on the other side yet on the picture that you posted with green chips on the wrong side they are on both lanes while they should only be on one of them due to earlier splitter.

I did a full cleanup of all transport lines in this area and left it running for at least 10 minutes and there were no cases of items landing on the wrong side. Was this build placed as a single blueprint stamp, or were you stamping that blueprint multiple times? I am suspecting it could be that this setup was stamped multiple times and because of 134255 some of the splitters got their items swapped between output lanes.

[SoYman]

I appreciate the quick response. That is a good observation, some circuits have landed on the wrong side indeed.
I have definitely iterated on this build a lot and the typical way to apply any small changes is to copy a larger rectangle, make the edits, delete what needs to be changed in the original build and paste the larger rectangle. This could have triggered the bug in 134255, although I'm not using many underground belts, in contrast to the observation from @Arcus.

IFAICT 134255 only swaps items between sides on a single belt, but in my case, the items were swapped between two belts inside a splitter. You obviously understand the underlying mechanics better than I and there might be some way—like the erroneous swap occurring inside the splitter—for the items to swap between belts.

It's very plausible that it's a duplicate of 134255. I defer to your judgement on deciding if it is.

Side note: I realised that that filter is completely redundant anyway but a bug is a bug.

[boskid]

It does not flip items on underground belt, but it flips items on a splitter. Result of pasting a flipped blueprint in 2.1.9 on top of a splitter which has items already past the filter results it items swapping lanes which is what was also fixed by 134255

07-10-2026, 16-03-17.png (127.86 KiB) Viewed 311 times

[boskid]

I am going to classify this as a duplicate of 134255 because it has high chance of being a duplicate while not showing any signs of splitters failure when not actively building.

[SoYman]

That makes sense. Thanks for looking into this! I'll make sure to report back if this isn't resolved in 2.1.10