Sushi thread

[hardy fable]

I just accidentally learned that you can flip a splitter with R

[foggy junco]

u mean rotate?

[hardy fable]

flip, I finally got the right term 😅

oh it's actually a 180° rotation

I wonder if we can do a mirror flip then

[foggy junco]

those things still look better than the solar bp i once designed though

gotta admit

[pulsar lance]

how would you tell the circuit to get a ratio of pi?
|| Right, as you can't enter all digits you need to first build a machine which calculates pi||
|| Then measure time and throughput and activate the belt for time/throughput<pi||
|| But afterall the correctness of pi you can calculate with is limited by the number of combinators (-memory) you use, which means: You only approximated it ||

[plain hare]

Hehe you could approximate pretty close if you really wanted looks like

Shit you could maybe flow limit to an approximation of pi with splitters I think

[foggy junco]

Guys

Dont give me ideas

Now i need an infinite sum that converges to some multiple of pi with only even terms

Hmm

[plain hare]

1 - 1/3 + 1/5 - 1/7 + 1/9 -… = pi / 4

[foggy junco]

[foggy junco]

Yeah but good luck with odd terms

[plain hare]

You could but it would suck

[foggy junco]

[foggy junco]

Though this has odd ones too

[hardy fable]

factorio is turing complete, thus you can compute as many decimals of pi as you want

[pulsar lance]

Yeah, but memory limited

[hardy fable]

as everything

[distant gorge]

I like how you don’t specify beyond Factorio, (because really practically every subsystem is)

Circuits, trains, belts, pipes

[hardy fable]

well, since even cities skyline is turing complete... XD

[distant gorge]

“Cities Skylines is Turing complete” is pretty vague, which specific thing do you mean?

It’s odd to me to call an entire game Turing complete, usually you refer to the specific mechanic that is

[hardy fable]

fine

then "cities skyline poop is turing complete"

https://medium.com/@balidani/cities-skylines-is-turing-complete-e5ccf75d1c3a

[distant gorge]

Ah, lol

Makes sense

[plain hare]

@hardy fable I was hoping you could help with a sushi pipe thing. I was wanting to unload multiple fluids from the same rail line in a compact space and I think it could be good here. You think that’s possible?

[timid cosmos]

Someone made a functional sushi fluid loader a while back

i don't rmemeber who though

[timid cosmos]

The biggest issue is obviously clearing the pipes, I've had the thought that every fluid would need a main storage tank and an overflow handling tank, and I've also thought a global check reading storage tank might be needed?

You have your main tank for each fluid that you dont read and you don't care about, that pumps into each of your respective handling tanks which you do read until <X. When the train comes, I imagine you're using LTN to output the fluid signal you're calling for? That loads the train by pumping the right fluid into the shared loading pipes. You also read if a train is there, when it is gone T=0, another pump turns on to suck everything out of the loading area shared pipes, and stashes it in a single storage tank, whose output signal activates the respective handling tank secondary input pump, I would even add another 4 second timer to it somehow or something, so the pipes are certainly cleared once the tank signal is lost. I'm not ingame so this is all theory crafted but I'm pretty sure it would work

You could store the trash fluid storage tank fluid signal in a memory cell which resets when T>0, skip any timer crap

The trash storage tank idea could probably be circumvented entirely if you're clever, but my design would have one just to be sure

Also I just read you want to unload multiple fluids, not load multiple fluids, so that theory crafting is pretty much useless for you

Unloading seems like it would be a lot easier

Since you can just used "Read Stopped Train" signal to turn on the correct pump to suck the right fluid out of the shared unloading pipe, you don't need LTN to output the fluid you need

as for making sure the pipe clears, you just use a larger buffer than the amount of fluid you're calling for

However I could foresee the memory cell idea being useful here again, to clear any potential fluid after the Stopped Train signal goes away, for certain

[hardy fable]

yes it is absolutely possible, I did it multiple times
the first one was using a lot of combinators but the last one barely use any

#quick-questions message

when you build the filter (the second pic) with bots, remove the top-most vertical pumps and rebuild them, to set the priority
if you build it by hand then just build those last
on the bottom-most pumps the condition is "enabled if anything < 24600", which will make the pumps right above bug (they won't empty their internal storage even if the tank isn't full) but it will work perfectly fine

[hardy fable]

this is definitely how my first design (using combinators) works, the hardest part was being able to repeat a signal after it went off

[red tendon]

I can't reach the first message lol

[hardy fable]

not that having this many pumps is not required

[plain hare]

Any way to auto prime the pumps loading fluid tanks with fluid barrels?

[hardy fable]

a way to what?

ah the last tanks

[plain hare]

Ye

[hardy fable]

just send a train with the top fluid first

then the second, etc until the bottom one last

since it's not filtering out at first it will unload on the first tank it can

[plain hare]

That’s an option for sure but it’s basically manual… would be nice to have this get built by robot somehow

Maybe circuiting the pumps leading south to disable until some heavy oil is in

Could put 1 barrel assembler on each pipe connection down there… hm

[hardy fable]

I see, I'm gonna have a look rn

[hardy fable]

maybe a break-away bootstrap like this?

[plain hare]

What exactly forces the filtering pumps from the sushi pipe from taking all the fluid instead of letting some go back down?

Or does the infinity pipe represent that we should loop the pipe around again?

[hardy fable]

by "go back up" you rather mean "partially continue going down the line"?

[plain hare]

Yeah early for me meant down

[hardy fable]

the infinity pipes are just the rest of your factory, tho the lowest is useless here

[plain hare]

So like. How come no heavy oil ever ends up in the pipe segment before the petro gas tank? What if the southern pumps took priority over the eastern

[hardy fable]

the fact that you have a single pipe with one pump in and one pump out

[plain hare]

Well it has 2 pumps out - how does it know to go east and not south?

[hardy fable]

the single pump in can't have a higher throughput than one pump out, so the game just give everything to the one that has been built last

so you need to make sure of the build order

[plain hare]

Ah so it’s build order priority I was scared of that answer

That makes getting bots building this even harder hmm

[hardy fable]

when using bots I just cut and paste the row

[plain hare]

Tru that’s pretty easy at least

What if you did loop the pipe like we loop sushi belts?

[hardy fable]

you still need a filter

[plain hare]

Oh I guess none would make it east and it might loop endlessly through the south

[hardy fable]

with sushi belts it's easy since we have filter splitters (btw when they weren't a thing back in the days, I was making my sushi belts using filter inserters 😅 )

[plain hare]

These sushi pipes are kinda cursed being build order dependent lol

I’ll have to decide if I want to use it or not

[hardy fable]

september 2017, that's a rare piece of history right there

[timid cosmos]

I've had thoughts on builds that I think would work, but I've not really considered build order, that seems unnecessary when we have access to the circuit network doesn't it?

[plain hare]

Yeah I think you could use a different approach potentially for sure

[hardy fable]

feel free to experiment, I haven't found another way to do it yet but there might be, who knows

oh wait

@plain hare if your goal is just to have a blueprintable universal fluid wagon unloader, you can use my combinator one

this one doesn't require any initial condition

or build order

[plain hare]

?

[hardy fable]

the main idea is to route the fluid depending on which one it is
and then once the buffer tanks are empty there's a big circuit that repeat the signal for a bit of a time so all the pipes are empty when the next train comes in

[hardy fable]

yes?

[plain hare]

That statement seems at odds with this one

[hardy fable]

they don't work the same way

[plain hare]

I thought we only had 1 sushi pipe build posted so far, and it’s build order dependent?

[hardy fable]

well well, depends on your definition really

I had this one lying around for some time but it's not really a sushi pipe since you can't have 2 fluids at the same time

[timid cosmos]

Another dude had an interesting design for a sushi loader that looked really nice

[hardy fable]

like it's one fluid, everything gets emptied, then the other fluid is loaded

it works for unloading stations but you won't run a refinery using only 2 pipes using that setup XD

[hardy fable]

with more than 2 fluids and max fluid throughput? sounds interesting

[timid cosmos]

I dunno about max fluid throughput, but he was running all fluids in the game through it, it seemed

[hardy fable]

anyway, the combinatorless filters fluids out using pipe priorities (build-order dependent) and pump content
the combinator one just routes the fluids depending on what's inside the storage and runs for long enough so every pipes are empty before the next fluid comes in the system

[hardy fable]

this one? #quick-questions message

[timid cosmos]

No, I think this player had a white name, but I don't remember what it was

[timid cosmos]

#blueprint-designing message

found it

I remember now he used a cargo wagon with a fluid barrel to tell the pumps which fluid to load

No clue how it works

[hardy fable]

oh but that's a loader

lmao he's using an entire wagon just to tell which fluid to load XD

one thing I'd do is to not fully unload the fluid wagon

[timid cosmos]

I think its a nice solution, I wouldn't know another way to select the right fluid without that, or LTN, or using the Train ID which is no fun

[red tendon]

i want to have stone bricks mixed with the iron ore any help ?

[hardy fable]

just to make sure we are not facing a XY problem, why?

[red tendon]

i cant put a extra belt for that

[hardy fable]

but stone is not required to make concrete

[hardy fable]

that's iron ore

[red tendon]

wait

[hardy fable]

you just need iron ore + bricks

[red tendon]

ooooooo

[hardy fable]

:D

[red tendon]

ty 🙂

[hardy fable]

np

[hardy fable]

@plain hare you were right for both 😄
#930147588160782436 message
#930147588160782436 message

no initial condition, rightmost tank is the input

[plain hare]

That’s what I’m talking about!!! My man!

[hardy fable]

the downside effect is that you have to use 1 pipe per pump

I haven't tested with more but I can see how it might become an issue

[plain hare]

Hell yea you I think you could definitely strap this up to some kinda compact fluid unloading station

[hardy fable]

and it broke in the most unexpected manner XD

the led is green

yet it doesn't pump lmao

[plain hare]

Now that’s odd

[hardy fable]

oh no

it depends on the build order as well 😭

[plain hare]

I’m guessing this happened then

Maybe you should time the pumps

[hardy fable]

which is obvious now that I think about it, it's like my combinatorless sushi pipe but with tanks intead of pipes

[plain hare]

Clock them like inserters

Yeah leave the extraction / filtration pumps always on and then cycle the looping pumps to be sometimes on!

That should work I think?

[hardy fable]

hmmmm

it just goes blink blonk

I don't even think it can ever work because 0.0 (or even 0.2) fluid outputs nothing in the circuit network

which is not the case for trains, weirdly enough

[plain hare]

Hmmm I must be misunderstanding something why do we need to measure the fluid tanks for this one?

[hardy fable]

the logic is if [fluid] > 0 then enable pump

[plain hare]

Hmmm damn you fluids

[hardy fable]

I have achieved sushi pipe lmao

[desert tide]

just as long as heavy oil doesn't end up in that last pipe segment

[neat idol]

Building anything "clever" with fluids is really pain in the ass thanks to this ":D"

• https://katiska.dy.fi/n/temp/factorio/examples/fluid-boxes-build-order.mp4
• https://katiska.dy.fi/n/temp/factorio/examples/fluid-boxes.mp4

(I even wrote a simple simulator when I was playing around with fluids: https://katiska.dy.fi/n/temp/factorio/code/flow.py https://replit.com/@warbaque/fluid-flow#main.py)

[hardy fable]

cool stuff 👌

[clear latch]

I made a circuit based throttler for moving average sushi. Moving average >that I used earlier< with no control beyond read belt pulse/disable belt will flood the belt until it finishes averaging out. For slower timescales like 3600 ticks/1 minute that can be a fair amount of items. I used to force a physical loopback with a splitter + disabled output but that required waiting a whole minute for it to "spin up".

This version checks the input belt for an item and then pulses that item's CC quantity to the MA with an edge detector to temporarily fake item satisfaction. In just a few seconds that decays and the real items start coming out. Any change to the input belt or CC is detected as a rising/falling edge and updated in the MA.

[halcyon wing]

I had a very similar idea, using a single belt to read the contents of the bus and doing some logic to control based on that.

[clear latch]

In my case the single belt would also be the bus 😄. So requests like 18, 36, 72, items/minute are common to represent whole vanilla belts

[halcyon wing]

Heh, this is for Space Exploration where there's a lot of weird items required.

Throughput requirements are relatively low.

[timid cosmos]

In my Sushi Village factory I read about 20 individual belt pieces spread around the whole factory and took their contents over ten seconds and divided by 20 for a factory-wide average and compared that against constant signals

but that factory had every single vanilla item necessary for progression on it

it was really slow to use sushi for the whole factory, but I learned that the item-over-time sushi is very robust because it repairs itself if players take items off the belt.

[halcyon wing]

Oh no.

I'm looking at mods and there's a lot of useful circuit mods.

Recipe as combinator. Satistics combinators.

I'm trying to figure out how I will set my sushi belt needs.

Recipe as combinator would let me use vanilla entities to transmit requests along power poles (easy enough to setup)

Satistic combinators could let me add items as they're consumed, which will reliably be guaranteed to be on one bus.

[clear latch]

Tested my new control circuit on a full set of stacked ore (x50) smelters. Each ore has its own flow-limited loop feeding back into the train chest; each output feeds into the main bus loop. Smelter power poles allow reading and disabling belts on opposite sides from one circuit. A few dummy variables separate the same ore with the same train stops used for different purposes (e.g. iron for steel vs for plate)

[clear latch]

A few months back I made a sushi item picker circuit. Each inserter read its own belt and picked faster due to never being stuck on unavailable items. CC output 1 + belt output 1 = 2 = set filter. Read: Pulse and Each = 2 saved an extra decider per inserter:
#930147588160782436 message
But if an item appeared on both belt lanes simultaneously, Read: Pulse = 2 instead of 1. This looked identical to CC output 1 + belt output 1 and eventually jammed chests.

I didn't want to sideload all items or add deciders. So I direct wired inserters to belts (no tick delays) and *-8 to cancel out filters, then tested Hold because the new circuit didn't use belt output anymore. Basically faster and always 6 combinators even with many inserters. It picked anything so I added a dynamic blacklist of items not in the CC. I also added failsafe wiring in case combinators were built after inserters. Missing signals disable input belts and force a splitter loopback upstream. Only my intersections have enough inserters to save combinators but it was fun figuring all this out

TL:DR; same flow rate and CC for all variants below, items left are out of 231 total

[normal gulch]

you beat me to it, i wanted to design smth like that

hm, is there recycling? @clear latch

I might not be 100% understanding what it is though

[brisk carbon]

Hi I'd like to scale up my science production. However I'm not able to get my second "feeder" to run without deadlocking my belt. As soon as I activate the top belt my belt get's literally flooded within 20 seconds

[hardy fable]

of course, there's no recycling

you're just shoving items into the loop it seems

[clear latch]

Important detail to prevent jamming. Has to be read in addition to all the inputs when it loops back

[brisk carbon]

what do you mean by "recycling?

[hardy fable]

from experience there is no point in having 2 "feeders" per loop anyway, just make 2 loops

[normal gulch]

a loopback, I misunderstood what its for.. still dont really know is it building a specific thing?

[hardy fable]

something that will eat away all items in any case

thus preventing jamming

[clear latch]

No the loopback and that one wired belt is blueprinted into the circuit because it's always necessary in every implementation of it I use

[brisk carbon]

erm doesn't the inserter count as recycler?

[hardy fable]

the one that goes into the lab?

[brisk carbon]

jup I red wired them all

[hardy fable]

no

a recycler eats everything all the time

a lab will only consume what it needs depending on the research time per cycle needed and if a research is currently active

[brisk carbon]

just like a green inserter followed by an active provider chest?

[clear latch]

This is what my bus of stacked items (1 stacked item: 50 unstacked) looks like (top control circuit, big loop going around). But the new circuit is for sushi rails to help speed up construction - one circuit per segment and a couple of inserters in every direction of belt flow/adjacent roboport construction overlap - so maybe 8 inserters for a 4-way, 6 for 3-way, 4 for 1-way, etc.

[hardy fable]

might work, that's somewhat how I was doing it before the game had filter splitters

in red you have the recycler, in purple the mixer and in green the prioritizer

all sushi belts require those 3 components

[clear latch]

Splitters with input priority have more throughput but you can disable inserters if measuring amount of items on belt loop or flowrate: #930147588160782436 message

[brisk carbon]

I see yeah my approach was like get an constant combinator and set it to valu X and the belt will fill it self up until the threshold has been reached.

and it did work however since I tried to get another feeder in for more throughput my counter always goes into the negative values which I don't get at all.

[plain hare]

whats the throughput desired for your setup?

theres an easy no combinator one thats also cheap on resources for early game

[brisk carbon]

I want roughly 160 items on the belt

however my 12 labs are so hungry that my other 12 labs won't get any science at all

[normal gulch]

so its building sushi rails got it

is it a mod that stacks items?

or am I misunderstanding that.. im thinking its 50x items in 1 slot on a belt

[clear latch]

Deadlock's Stacking Beltboxes & Compact Loaders (main mod), Deadlock's Stacked Mining (drag selection tool over ore), Deadlock's Experimental Stacking (direct smelting of stacked ore into stacked plate). So yeah 1 item literally equals 50 when unstacked, super low throughput

[brisk carbon]

let me guess I'm kinda stupid I could just "split" the signal from those combinators and be feeding those feeders right?

[normal gulch]

why do u do that though?

[clear latch]

1 belt bus (er technically 2 belt profile)

[normal gulch]

ohhh 👀

[clear latch]

I imagine part of the problem is measuring item count rather than flowrate. Flowrate can be a fairly large number but belts can't physically hold all that many at once (8 items in straights, 7 in corners iirc) so rule out the obvious possibility that you're requesting too many (because it looks like thousands from that screenshot)

[brisk carbon]

nope I was requesting just 10 items.

and I did reset both counters beforehand. And it took roughly 20 seconds to literally shove ~200 items in

[clear latch]

Maybe you have two independent circuit controls and when one side of inserters swings it decrements the other counter even though the items are not from there 🤷‍♂️ . I'd try simplifying things with a single counter and wiring everything together as one network. If nothing works you could link a bp string of it for someone to look at

[brisk carbon]

ok this was my other attempt and the results where again devastating

https://pastebin.com/gQmPUBKc

[clear latch]

Pastebin deletes bp strings, try https://factoriobin.com/ or just paste it here or upload as text file

[brisk carbon]

https://factoriobin.com/post/aDjMJmxSuWQLmhK_-EXPIRES

thanks I didn't know about that

and I just droped 5 red bottles and this is the result

I just discovered something. For some unknown reasons my belt readers subtract the counter instead of adding the value

[brisk carbon]

I found a solution

I had to wire a second circuit which didn't connect to the inserter

[clear latch]

Yeah the minus was being counted because the two networks were bridged

[halcyon wing]

Is there a way to read the biggest signal on a circuit network?

My sushi cleaner isn't picking up something that is way overfilled.

I tried being creative and having one machine feed from and output to the same belt.

I thought the cleaner would just sweep off any excess but it didn't. 😦

[pulsar lance]

Well yes, with enough combinator stuff, there are versions of max signal finders with different advantages and disadvantages (space, efficiency, relying on recursion)

Not sure about your altogether problem though

[halcyon wing]

I stopped outputting to the belt but...

There's still a lot of these green cards.

[plain hare]

What control mechanism are you using here?

I see many combinators :D.

[halcyon wing]

I use a timer. Each cell remembers 1 minute worth of items, then resets after 5 minutes (with 5 cells)

I take requests - seen and insert that onto the belt.

If requests - seen < -10, it's sent to the cleaner inserter.

I added a second one for requests - seen < -1000

[plain hare]

Definitely a new approach from what I’ve seen.

How do you ensure the belt doesn’t overfill? Or does it not really matter somehow?

[halcyon wing]

I haven't figured that out yet.

I'm using a mod that creates a constant combinator from a specific crafting machine's recipe (vanilla behavior, just a handy shortcut) and I wire that up to make requests.

Since each machine isn't requesting more space than it's footprint on the belt, it's not a problem.

[plain hare]

Could the total sum of machines request more than the belt has capacity for?

And also you’re tracking exact counts of items, right? Not their item per second rating on the belt?

[halcyon wing]

Yeah, I'm counting each item as it passes by.

[pulsar lance]

I think someone made a sushi factory with measuring at 10 random places for limited time and compares that to „request“ to decide which items need to get produced could be (though without the request per assembler stuff)

[plain hare]

Yeah that was my boi Tangled Up in Bluegrass sushi village

I helped them a little in the early stages of that planning

I think the rate limited belts sort of inspired them to use rates in a different way

I think that was a bit different but I’m not 100%

[halcyon wing]

Now I need to figure out how to get the dozens of items to the sushi belt.

[plain hare]

More sushi belts!

[halcyon wing]

I'm using bots right now and that works.

Space exploration is just weird because it just requires so much random crap everywhere.

Literally the perfect use case for a sushi belt.

[plain hare]

You know I love sushi belts so I hate to ask but

Why not just have the bots deliver to the machine if you already have them?

[halcyon wing]

1stly, bots crash in this mod.

2ndly, if I can get a working design I can tackle some of the harder mod packs.

My last attempt was 100% bots and I just treated the crashed bots as a cost of being in space.

[clear latch]

Definitely did not think this one through...I think I'll actually overhaul the belts so I can keep the central combinators, especially the CCs. It would be free (no insulators just change wire color to duplicate those without redundant requests if the inserters weren't already double wired