[2.0.33] Constant combinator signal of 2^32+1 does not overflow in GUI

[Akino_Ovsky]

In a constant combinator you can input 2^32+1 and end up getting an output value of 4294967297

1: Open constant combinator
2: make new signal
3: set value to 2^32+1
4: get a value of that, without it normally overflowing to 1

[Akino_Ovsky]

Okay, with a bit more checking you cannot do this with decider or arithmetic combinators in any way, nor does the Constant combinator actually output the gigantic invalid signal value, once you close and reopen it, it will reset the signal to the correct value which would be 1.

It will also not do it with anything like 2^32+2 which will correctly overflow to 2 in the UI without having to close and reopen
I think I just found a weird edge case and this bug can I'm pretty sure be safely ignored/fixed with a single line of code

[Loewchen]

It seems to just be unreliable and any value that should overflow has a chance of being displayed as if its a 64bit number until the GUI is reopened instead, so an input of 1000000000000 or 1e12 both show sometimes as 1T. It only happens when there are no other signals set in the combinator though.

[Rseding91]

It looks like this is fixed in one of the most recent releases.

[protocol_1903]

This might've broken a feature (or bug idk) that some of us in the circuits community have been using for a while. If you set a constant combinator in one logistics group to INT_MAX or INT_MIN, then in a second group overflow it with a much smaller number (like 10-20) the combinator would correctly add the two numbers together and overflow them on the output. I've been using this as a cheap way to limit assembler ingredients (combined with read contents and set filters). This has since been broken (in 2.0.44) and instead it just sets it to INT_MIN or INT_MAX without the overflow. I'd like it returned, if possible. Should I create a bug report for it?

[Loewchen]

...

This has nothing to do with the issue discussed here as it was limited to the GUI only. If you have a problem in the latest version make a dedicated report following
3638 instead please.