[0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Posted: Sun Feb 23, 2020 5:05 am
by Moleculor
There's 24k listed in the contents of this Heavy Oil fluid tank.
The signal it's outputting is showing on the Decider Combinator as 23k.
I can't tell if it's a display/UI bug, or what, but it's weird enough that I thought I'd mention it.
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Posted: Sun Feb 23, 2020 9:12 am
by Blacky007
add a digital numbersystem to see the exakt value https://pastebin.com/nbvkrCdZ
maybe it switched from 24k to 23k in the moment you switch between the tank and combinator?
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
It's because the circuit network rounds down (even 23.999) and the devs will not change it to a useful behaviour.
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Posted: Sun Feb 23, 2020 2:22 pm
by Loewchen
The UI rounds symmetrical while signals are rounded down, so this is working correctly.
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Posted: Sun Feb 23, 2020 2:47 pm
by Oktokolo
Loewchen wrote: ↑Sun Feb 23, 2020 2:22 pm
The UI rounds symmetrical while signals are rounded down, so this is working correctly.
Looks like a user experience bug to me. Circuit network UI and fluid tank UI (and all the other UIs too) should use the same rounding algorithm for number display.
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Loewchen wrote: ↑Sun Feb 23, 2020 2:22 pm
The UI rounds symmetrical while signals are rounded down, so this is working correctly.
Looks like a user experience bug to me. Circuit network UI and fluid tank UI (and all the other UIs too) should use the same rounding algorithm for number display.
This goes as far back as fluids exist, every possible permutation of what to round in what way existed at some point and all of them were "obviously a bug because...".
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Loewchen wrote: ↑Sun Feb 23, 2020 2:22 pm
The UI rounds symmetrical while signals are rounded down, so this is working correctly.
Looks like a user experience bug to me. Circuit network UI and fluid tank UI (and all the other UIs too) should use the same rounding algorithm for number display.
This goes as far back as fluids exist, every possible permutation of what to round in what way existed at some point and all of them were "obviously a bug because...".
It may very well be the case that there is no right way to round these numbers (there definitely is no lossless one ).
But whatever rounding algrorithm is used should be the only one used in all the places where these numbers are displayed.
Rounding towards nearest integer is okay. Rounding towards zero is okay too. Just don't use both for the same number depending on where the player looks for it.
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Posted: Mon Feb 24, 2020 2:08 am
by Moleculor
Blacky007 wrote: ↑Sun Feb 23, 2020 9:12 am
maybe it switched from 24k to 23k in the moment you switch between the tank and combinator?
You mean the five times I switched between them? (And just now, when I go look at them again in game?)
Loewchen wrote: ↑Sun Feb 23, 2020 2:22 pm
The UI rounds symmetrical while signals are rounded down, so this is working correctly.
I'm not even sure what "rounds symmetrical" means, but how do I know which number to trust? Which number is more accurate? Which number is going to affect the circuits? I feel very confused by this version of "correctly".
If it's going to round, why not round it the same way everywhere?
Re: [0.18.8] Mismatch in fluid tank signal value and fluid tank contents?
Loewchen wrote: ↑Sun Feb 23, 2020 2:22 pm
The UI rounds symmetrical while signals are rounded down, so this is working correctly.
I'm not even sure what "rounds symmetrical" means, but how do I know which number to trust? Which number is more accurate? Which number is going to affect the circuits? I feel very confused by this version of "correctly".
If it's going to round, why not round it the same way everywhere?
"Correctly" as in the way the dev decided it should work.
You can not get the exact signal value without commands or mods, but you can tell that it is between 23500 and 23999.
It does not round the same for both because that would either have meant to break every existing circuit logic that needs the current behavior or to not fix a bug way more confusing than the issue reported here.
).