[Sharing] My Journey Creating AlteryxBigInt Custom Functions

Inspiration

During the Euleryx series hosted by @Pilsner , I encountered many challenges that required computing with BigInt (numbers exceeding 100 digits—something most built‑in functions struggle with).

Recently, seeing Pilsner using a custom IsPrime function created by @gawa to speed up runtime motivated me to explore building my own custom functions for handling BigInt.

The Functions

To support large‑number computations inside Alteryx, I developed a set of custom BigInt functions that return results as strings:

BigInt_Add(a, b)
BigInt_Sub(a, b)
BigInt_Mul(a, b)
BigInt_Div(a, b)
BigInt_Mod(a, b)
BigInt_Pow(a, b)

I’ve attached the .xml and .dll files below. you can add these file to the location mention in Alteryx Help

WARNING: Only Tested in a simple workflow, PLEASE TEST before use in PRODUCTION workflow

Below is my journey to build these custom function:

First Attempt: Duplication

My initial approach followed two key resources:

I installed Visual Studio, configured the environment, copied example code, and compiled the DLL and XML files.

But it didn’t work.

After trial and error, I discovered the issue, I forgot to delete some unused header .h and source code.cpp files, which caused conflicts during compilation.

Lesson learned.

Trying to Create My Own Function

Although I had experience with Go and Python, I was new to C++.
I only watched a few TikTok videos about C++, which surprisingly helped later.

I only option is to create BigInt function with the help of AI. Start with BigInt_Add

But it didn’t work.

Since I had no idea at all, I decided to implement IsPrime from scratch, because the function is smaller and simpler.

Duplicate Creation, Now With Understanding

This time, I could finally follow the flow of the code and begin to understand how everything worked under the hood.

The good news? I started seeing different errors—this time without any Alteryx error messages.

That was a positive sign. It meant the function was compiling correctly, just the function is tell Alteryx the result is error.

After going back to both article and reading them more carefully, I uncovered the key detail I had missed earlier.

In normal C/C++:

0 = success
non-zero = error

In Alteryx custom functions:

1 = success
0 = error

AI-generated code followed standard C++ conventions, which caused Alteryx to treat valid results as error.

After reversing the return values, It worked perfectly.

Testing the BigInt Functions

With the return values fixed, I applied the same change to my BigInt functions.

They worked inside the Formula tool. Time to stress-test with something heavy: Fibonacci(1000).

Using BigInt_Add in a Multi‑Row formula:

IF [RowCount] <= 2 THEN 
  "1" 
ELSE 
  BigInt_Add([Row-1:fib], [Row-2:fib]) 
ENDIF

But I got an error, with no error message again, test with the IsPrime function produced similar issues.

And using hard code value, both work fine.

So the only reason I can test think of is NULL or empty inputs.

Requesting AI to accept the Null or Empty Inputs. and Bingo, problem solved.

Final Creation

With this working structure, I asked AI to help generate the remaining functions:

Subtraction
Multiplication
Division
Modulo
Power

All of them now run correctly in both Formula and Multi-Row tools.

Conclusion

Building custom BigInt functions for Alteryx was a challenging but rewarding experience, especially for someone new to C++. The process helped me understand the Alteryx SDK, C++ return conventions, and the importance of handling NULL inputs.

I hope it help you to build your own custom function. Join us in Eulteryx.

BigInt.zip

Custom Tools

Installation

Professional (Edition)

Accepted answers

All comments

Pilsner

@AncientPandaman , these functions are brilliant! I've already downloaded them - looking forward to trying them out later. Thanks for including the write-up of how you created them, too. I've been hesitant to try custom formulas in the past, but glad to know you found it was worth the time investment, Ill have to try and create one myself!

Gawa

@AncientPandaman Great work! LLM helps us a lot learn about C++ syntax and trouble shooting. I admit I almost count on Gemini😁

What comes to my mind when I use you 'BigInt' custom functions is...it could help not only huge integer calculation but also even precise int64 calculation.

Unfortunately Alteryx Designer can cause precision error for particular calculation of int64(thanks to @Qiu for letting me know this). The below table shows example of precision error ;Pow(7,19), Pow(7,20), and Pow(3,34). Needless to tell Euleryx persons(!), power of odd number is always odd number, however Pow(7,19), Pow(7,20), and Pow(3,34) returns even number ending with even numbers, 4,0,and 8.

On the other hand, by using big_int custom functions, they return precise result.

In our company's business domain, we don't encounter this kind of huge int64 so not a big problem, but in financial use case(maybe @Yoshiro_Fujimori ?), they could. In that case, big_int custom functions will help validation of calculation results of int64 to avoid precision error.

Anyway, thank you for sharing fun functions!

Precision_Error_INT64.yxmd

Qiu

@AncientPandaman

Thank you very much for sharing this one.

this will make the Euler Project problems much easier.

@gawa
Good point.

AncientPandaman

@gawa @Qiu interesting information. I thought still accurate as long as it show the number.

I test for double as well. but minus behavior is more interesting. It seem like convert to Int64 and do the minus.

and what more interesting is the BigInt_Sub create a different behavior.

it treat the double as 0 (due to out of Int64 range). but treat Int64 as null.

Screenshot 2026-02-26 112927.png

Precision_Error_INT64 vs Double vs BigInt.yxmd

Qiu

@AncientPandaman
Custom Functions have been really helpful. Thank you very much.
I am wondering if you can be kind to create the Prime Factor functions including Large Int which will also be very much useful, especially for the Euler Projects Problem solving. 😁

AncientPandaman

@Qiu I’m not planning to create new function.

My current focus on improve these functions to

1) deal with different situation (like decimal or scientific format); and

2) adding error message (like div/0 mod 0)

Current roadblock is the error message have different behavior with same config.
1. trigger error that stop workflow run (if it refer to a field, it error no matter the value)

2. just a warning message

Beside, I think it against the purpose of the BASE‑A method.
If it is non-BASE-A, it might be better to use the Python tool instead, since it easier to use and setup for both designer and the server.

Qiu

@AncientPandaman
Thank you for the update.
I perfectly understand your focus now and looking forward to your new update.

Those functions have been really helpful.

Quick Links

This months top contributors

atcodedog05 19598

Qiu 15878

binu_acs 15708

MarqueeCrew 13708

apathetichell 13703