[Dominik][0.17.63] Shared fluid system with separate pipes

[boskid]

I am not sure how this happend, but i have save file with custom "control.lua" that performs bunch of randomized actions with fluid systems. At some point (tick 74430) two disjoint fluid networks managed to connect (both locked by sulfuric-acid and share same system ID) when they should not be (there is cutter underground pipe present).

Steps to reproduce:

1. Open fluid_fuzzer_zip.zip

   fluid_fuzzer_zip.zip

   (133.49 KiB) Downloaded 107 times

2. Increase speed in editor to x64
3. Wait 18 seconds
4. At tick 74430 my custom script should print error message and stop itself

   

   74252-disjoint-network-with-shared-filter.png (116.45 KiB) Viewed 2377 times

5. Fluid network on right is now connected to sulfuric acid even in presence of cutter underground pipe that should prevent it from connecting to filter

After this issue happend, is it possible to remove many pipes, change assembler to petrol unbarrel and all isolated pipes will also change its fluid filter (proves they share same fluid system):

74252-state-is-persistent-and-filter-is-truly-shared.png (102.43 KiB) Viewed 2377 times

-- edit:
- Is it possible to extend consistency check onto this case? To check if all fluidboxes in one fluid system are in fact somehow connected?
- Is there a switch (something like toggle-heavy-mode) that would run consistency check every tick? "toggle-consistency-mode" ?

[boskid]

[0.17.64] Manual reproduction

74252-manual-network-split.gif (933.57 KiB) Viewed 2248 times

Issue looks to be with placing pipe-to-ground ghost that should split fluid system (solution 1) or that should be part of fluid system so its removal should perform full split network logic and create new fluid system.

Compact reproduction:

74252-manual-network-split-simple.gif (200.48 KiB) Viewed 2248 times

[Dominik]

This one is a good catch, a very realistic and also serious issue. Thanks.
It is quite certainly connected to the 74269. I will check it out now.