[0.18.8] Side load with 2 saturated half belts produces gaps

[darklich14]

Attached is a basic video of 2 saturated half belts merging to make one full belt but the full belt gets gaps as shown with show-transport-line-gaps in debug settings.

[Rseding91]

Thanks for the report. most likely you have some circular belt(s) and the update order is what's causing this. Without seeing the full factory I doubt there's any kind of bug here.

[darklich14]

Here is a preview gif that is larger than the attached mp4

[darklich14]

Thanks for the report. most likely you have some circular belt(s) and the update order is what's causing this. Without seeing the full factory I doubt there's any kind of bug here.

Definitely no circular belts. Here is the destination of the full belt.

pic

Screenshot from 2020-02-21 11-39-09.png (6.52 KiB) Viewed 6915 times

Screenshot from 2020-02-21 11-39-27.png (747.5 KiB) Viewed 6915 times

The belt terminates to dead end at the last lab module.

[Loewchen]

Post the save.

[darklich14]

Here you go: https://file.io/nNKMmp

Of course it's going to expire at some point.

I pinged the location immediately before save. I look forward to your analysis.

[Rseding91]

Circular belts, or super long belts as there's an upper limit for how long a belt segment gets before the game stops recursively updating them to avoid crashing the game.

[darklich14]

So what do you do to maintain compression over long distance?

[ptx0]

trains.

[eradicator]

So what do you do to maintain compression over long distance?

Loader->Chest->Loader (or inserter equivalent) should work.

[darklich14]

trains.

I can't disagree with this.

I wonder what the recursion limit is. That would be a useful number to keep in mind while designing.

[Rseding91]

trains.

I can't disagree with this.

I wonder what the recursion limit is. That would be a useful number to keep in mind while designing.

1000 merged groups.

[darklich14]

Please forgive me for what may be an already-answered question, but can you please define "merged group"?

[Gizan]

Experienced this issue.... the first time i noticed it was 0.18.6, but couldnt figure out why it was happening

[Gizan]

Switched your belt to the other side

Capture.PNG (429.33 KiB) Viewed 6671 times

, and it moved the gap to the other side of the belt.

What i am seeing, is when that side of the belt stops, when it starts moving again, it doesn't take the item from the sideload belt.

[Gizan]

This space science belt is 67 belts and 59 UG's long and it gaps with sideloading too

[boskid]

Link expired so i did my own reproduction. Note, that sideloading will split transport line after 2 tiles (some optimisation) and this is where gaps show on your gif. Transport line should update at most once per tick and when it is half full it may happen that another transport line will feed it an item activating it (it cannot update this tick because last item was already moved and this would create a gap - so it is placed placed in front of queue so it will be updated earlier in next tick) and at same time, transport line that is fed by given transport line may activate it because now there is free spot at end of transport line. Since inactive transport lines will not force update order, if that order of other transport lines was wrong during activation, gap will create at front or at the back of transport line being activated. From my point of view this is only a minor issue.

[darklich14]

How are we supposed to build designs based on the premise of saturated belts if we cannot saturate belts?

[Bilka]

How are we supposed to build designs based on the premise of saturated belts if we cannot saturate belts?

There are many other ways to compress belts besides sideloading (I personally never had any issue with sideloading producing gaps though). Splitters and inserters come to mind, perhaps also splitting the belt and sideloading with two ends (to allow a buffer that then fills any possible gaps). As a sidenote, before 0.16.25, splitters were the only reliable way to compress belts (https://factorio.com/blog/post/fff-231).

[netmand]

How are we supposed to build designs based on the premise of saturated belts if we cannot saturate belts?

every system has its limitations. Have you tried side-loading more than once? inserters to fill the gaps?