[kovarex] [0.16.1] Items on belt creeping forward at specific position

[AntiElitz]

Iron Plate behind Inserters "jump" even tho compression fails. This is not really something game breaking, but may give a hint to something not working as intended right there



save attached

[mh_]

I did some tests on this bug. The first blueprint below is an example.
Fill the chest with the items of the corresponding smart inserters. If you turn the "bug" constant combinator on, the bug will show (the green circuit is jumping, therefor the yellow belt cant fit on the belt). If you turn it off, it works as expected.

It also depends on the length of the looping back on the right. If you increase then length to the right by 1 or 2 the bug stays. If you increase it by 3 the bug disappears (see second blueprint). Can anyone confirm my findings?

This should be a bug in the belt improvements, as it only happens when there are items on the belt further down. I suppose it is some sort of fence-post-error, since my testing suggested this would be a jump by 1 pos. The changing the belt length changes if there is a corner between the items further down the belt. Hope this helps.

Code: Select all

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

[edit]Blueprint needed stacksize 1 on inserters[/edit]

[riking]

Is it possible that the creep was supposed to happen before the inserter, and a sign error (plus vs minus) is making it happen after?

[mh_]

Is it possible that the creep was supposed to happen before the inserter, and a sign error (plus vs minus) is making it happen after?

This does not explain the behaviour I observed, please try my blueprints to see for yourself. Also I think that NO item creep in ANY WAY should occur.
I would rather suggest, that this is a fence-post error https://en.wikipedia.org/wiki/Off-by-one_error. This could easily happen when you would ask yourself if the item start on the belt at position 8 and the item has a length of 9, where does the item end and where should I place the next?

Note that I suspect, that the "no more sideloading"-issue is directly related to this very bug.

[Aidiakapi]

I've just ran some tests with compressing belts with the circuit network, and this bug is preventing that from functioning too. It feels safe to assume that any combinator contraptions relying on the timings of belts are broken in 0.16, even those that don't sideload, or rely on compression.

This setup is supposed to mimic a yellow-belt smelting array, the boxes get a single iron plate every 3.5 seconds, as if they were a furnace. There are 24, so the throughput is high enough to compress a yellow belt.

Video for 0.16.4: https://youtu.be/oV1JBpOTp7E
Expected behavior in 0.15: https://youtu.be/QuAybTuNo_U

Having done some more testing, it seems that it happens when a track gets "reactivated" (goes from white to blue in the visualization), I made a recording with visualization on and the game time lowered:
https://youtu.be/ANP7Lswdt7k

Before someone feels the need to point it out, yes the build did leave a gap in 0.15, which comes from the synchronization, and can be mostly fixed by detaching the last inserter's wires, there's always the probability of it happening, but it becomes extremely low.

Edit A bunch of blueprints for all combinator controlled smelter setups that work perfectly in 0.15 (without relying on undergrounds ofc). Probably not working in 0.16 due to this bug. BPBook: https://pastebin.com/JcrRLAJh

[mh_]

To all who do this with the default 12-smelter (stone and steel) yellow belts setup (as hinted in the blueprint):
1 78 14 91 27 104 40 12 53 25 66 38 / 105
These will circumvent the bug, as it depends on the free lane length as described above.

The numbers for the 24-version should be (not tested):
1 123 119 31 27 149 145 57 53 175 171 83 79 201 197 109 105 47 13 155 131 53 39 161 / 210

[DaveMcW]

Confirmed it's still a problem in 0.16.12.

Here is a test map with two jumps, one furnace width to the left of the player, and one furnace width to the right.

[Gergely]

I had a feeling that something like this would happen eventually, after reading all those FFF-s about belt compression lately.

Consistency is nothing. Optimisation is everything! Optimisation is a way of life!!!

[mh_]

I had a feeling that something like this would happen eventually, after reading all those FFF-s about belt compression lately.

Consistency is nothing. Optimisation is everything! Optimisation is a way of life!!!

go away. experimental versions mean bugs. spread your schadenfreude somewhere else.

[kovarex]

This particular behaviour is only happening when the belt is a circle. There needs to point, where the circle connects to itself.
When I turn transport lines debug option, it actually confirms, that the loop is closed at this point exactly.
Is this a really that big of an issue?

[Zavian]

I can reproduce it in 0.16.15 in the belt-bug.zip save that DaveMcW posted in viewtopic.php?f=182&t=54693#p330081 . That is just a straight furnace line, not a loop.

Edit: fixed link

[mh_]

This particular behaviour is only happening when the belt is a circle. There needs to point, where the circle connects to itself.
When I turn transport lines debug option, it actually confirms, that the loop is closed at this point exactly.
Is this a really that big of an issue?

This does happen if the belt is not a loop. Feel free to use the supplied blueprint as a testcase. The jump occurs at the mentioned transport-line intersections, but it is dependent on what is happening further down the belt.

Code: Select all

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

[mh_]

Please enable the mod "Creative Mode (Fixed for 0.16)" for the following complete linear blueprint.

Code: Select all

0eNrVml9u4zYQxq9S8NkqJIqUbAP7sihQBCj61qdiIcgSnRDVP1CUsW6gA/QePVlPUtLKOopMK0OxsbsvSaxYP9HffDMcj/SMdkXHGsEribbPiGd11aLt78+o5Y9VWuhj8tgwtEVcshKtUJWW+hUrWCZFXfHMy7jIOi5Rv0K8ytlXtA36lQFw4EJ26siZMbzDexidifsvK8QqySVnwzpOL45J1ZU7JhT6fHYmWCr5gXllnTNvz78m5yPZE2uluk5TtwpUV3oRCu7FK3RUv2iv1zcBYxfwegYcuoDDGTBxAZMZMHUABzPcyIHrz3DjM3fPC8mEx6uWCfXH9UgRZTlldeXgItmxp/TAa6Hf8uLlRP0vP5+356KViZ2fdR7JVCdVEOpXZZOKVOqLoE+oH95QqRTip3x7RoH+8SgYq8au5znarvsv+v3DRxty4luana/MVSZ6TZFKhnQC1QcmBM9ZopaQ/aFW/id7ScqJcmsL5eKbKxdSN+Xi0SfWr4MQIGVWN40SYoGYmzNDirRqm1pIb8eKmXoRqsXkXAwfRhe/S2jgw6kxnBrAqRGciuFUCqeGFiYlt09vHLjm98SlBODSVjJWLDFpQCzUDG+vJnGslirHJ3JuAHKauhkrVSnc+QTu/AhODeHUGE7FcOoaTg3g1A3crf7NzYrxxtGsZGJW7APMOtHXyqjYhwsa3F5Q6lhLVbK/FZRGAEH3aSs9J1Xh26kP9j6G76bwhMIhGArPfUzAUHiZwvCSCq+oGF5R4Q0Kjme+1+hvzd6hVl68uIK5XYtMV4AXV3i3hu371eB9amjfr0KorwlWspx3pTfs2Gq/buqCXe8vsfEbObbufiFrDK27XwiVWHcWECq17iwg1Mi6s4BQY+vOAkJd25ZsCHRjW7IBUOJbeh/PWJ8EtqUaskJsu6lAoKFtqYZAie2mAoHCkwle8UhkW/Ih0IU7lbn8m3YqYj9R8t9SyXczXiIXzebHzpfIxnrOeTdtXYee5COnntR3v4kQGAfngftNBDMYW09t7hd61xEOnY5waHyDEQ61HzPeT2LXmSOZDh0p/sihIyXu98DMeUHd74GZwZHtnORuZnAemmD/Mt/eHoiUGt9WfjrjFkMVGtvO/u4XAtdBIMVL5lZO6q6db/KaE2fjfJPXyI1Ge/aL7p5SfMerk+LXtmkyaYrNlhirDPbA58ED3akgL3z04TcjA1sxfjYydCtqNqBg+dR+EXmb34E2nykIC8cw5qcBsGVEw4+O6MN/ENFfR4zQ98cPttgUBHBEwkUi0u9HxI2DiBQo4mt7krNMVU8xr2HwI7Ww4gvydb9pF2w4nycN4MV2o0p/J5vOAsnUVnGUT7x6HNjNMTlJnuxFXSa8UjC0laJjvUUhmbYShhCsrgRrCC4GRxO/06RcjTZdFm16t2j7/9doE3Bww/eCSwwdpil4kVXw/Ntn6qn+tUwzEtuCN4rxP3/9vSDKDwuDa45J9E7CGWI2G+OLBDX1vKagx4sm4KrrUH48Pba6HT3lukJFqvpldewXXrFU/PATK3WqKeU1YYVUprTDgHMdY5/QTRxHff8v77StIg==

[edit]Since the contents of the creative chests get fucked up, please make sure it contains the right items by copying the the working chests.[\edit]

[Bilka]

I was testing something else and also noticed this. The attached save contains 3 simple setups. Rotating one of the input loaders of the middle setup to stop the input and then rotating it back will show that the items creep foward in the middle of the belt before the splitter:

[nuhll]

I dont understand the picture, what do i see there?

edit: okay at 2. picture i saw it, then i also saw it at the first. lol.

[Aidiakapi]

This particular behaviour is only happening when the belt is a circle

As you can very clearly see on the video I linked, it happens on straight track. It seems to be when a transition from white to blue state happens.
12 seconds in, the item after the second set of inserters on the left jumps forward, then at 19 seconds, that same item jumps forward again.

[Monara]

This is not exclusive to loops.

[Notnaton]

Item creeps forward when its "loaded" to the next belt "section"...

https://www.youtube.com/watch?v=CnZzJmOQ7ww

[BeeFriend]

I'm seeing the effect with something I am working on, but much more pronounced.

[mh_]

I'm seeing the effect with something I am working on, but much more pronounced.

Would you be so kind as to do the effort to blueprint your layout? From the video I cannot see the creeping and also cannot rule out the circuit network being at fault.