[2.0.73] "Set output side" on splitters behaves incorrectly when combined with a filter

[Castar]

When setting the output side of a splitter through a circuit condition and also using a filter (either manually or also set through circuit conditions), the behavior when neither the "Left" or "Right" conditions match is different from when not using a filter, and not consistent with the truth table provided in this post - EDIT: I misread this and it is technically the table for the input side. I guess it is possible that the intended behavior for the output side differs from the input side, but that would then also be confusing and unclear in itself.

Steps to reproduce

1. Without a filter setting (behaves as expected)
   1. Place a splitter. Do not change any of its default settings.
   2. Connect the splitter with a wire to a constant combinator, but don't output anything in the combinator yet. Check "Set output side" on the splitter. As expected, nothing changes under these conditions, no output side is preferred.
   3. Set the constant combinator to output O = 1. The splitter will set output priority to the right side.
   4. Clear the output of the constant combinator (or set O = 0). The splitter will revert back to preferring no output side, i.e. the default splitter behavior (again, as expected).
2. With a manual filter setting
   1. Place a splitter. Manually set a filter for any item. The splitter will automatically change its Output priority to "Left", but it is perfectly possible to set it back to the middle setting, making it behave like a "default" splitter once again.
   2. Connect the splitter with a wire to a constant combinator, but don't output anything in the combinator yet. Check "Set output side" on the splitter. Nothing changes, no output side is preferred, and the filter setting remains what it was before. This is still as expected.
   3. Set the constant combinator to output O = 1. The splitter will set output priority to the right side, filtering the item from the filter setting.
   4. Clear the output of the constant combinator. The splitter will not revert to default (preferring neither side), but will instead be "stuck" with output priority "Right". Circuit conditions can cause it to switch the output priority to "Left", but nothing can change it back to no output side preference. This is inconsistent with the setting's behavior when not using a filter and inconsistent with the aforementioned truth table (EDIT: maybe. See edit at the top of this post). Also, the "system" is now in the exact same state as in step 2, yet the behavior is different.
3. With a circuit-controlled filter setting
   (this is not meaningfully different from the previous example, but is an ostensibly useful and thus likely common scenario in which players will encounter this bug)
   1. Place a splitter. Do not change any of its default settings.
   2. Connect the splitter with a wire to a constant combinator, but don't output anything in the combinator yet. Check "Set output side" and "Set filter" on the splitter. Change the "Left" setting under "Set output side" to "Iron ore > 0". With no input currently, the "Filter" field will change to show the "No item" filter, and no output side will be preferred (once again as expected, so far).
   3. Set the constant combinator to output "Iron ore = 1". The splitter will set output priority to the left side, with a filter on Iron ore.
   4. Clear the output of the constant combinator. The filter will correctly revert back to the "No items" filter, but once again it will not revert its output setting to default (preferring neither side), instead being "stuck" with output priority "Left". The same inconsistencies as described in the previous example apply here.

[eugenekay]

When setting the output side of a splitter through a circuit condition and also using a filter (either manually or also set through circuit conditions), the behavior when neither the "Left" or "Right" conditions match is different from when not using a filter, and not consistent with the truth table provided in this post.

The table in the linked post is for the behaviour of the input side.

[Castar]

Thanks for the heads-up, I misread that and have edited my post. The other inconsistencies remain, however. But also, part of the confusion and inconsistency is precisely because controlling the output side does not behave the same as controlling the input side, even though their manual controls are identical.