Is there a guarantee of encoded sequence uniqueness with differing salts? about hashids.net

[manigandham]

I haven't looked at the algorithm in a while but you can check it here: https://hashids.org/

Different salts with the same integer input will have different outputs. Different salts with any arbitrary input will have duplicate outputs at some point as far as I know, and you can test this by decoding encoded hashes with different salts. This should be rare though.

If you already have your categories set then you can do a quick test by generating a 100k values for each and see if any duplicates show up. If you have enough unique IDs to work with then you might not have to worry about it.

[ullmark]

I am not sure either tbh, but I would guess that the outputs can contain duplicate values in that scenario.

But why not just let salt remain the same and pass in the category as the first value and the "id" be the second. [1, 4] to generate a short id?

[ckpearson]

Thanks for the reply both, will take a look at generating a large number of IDs to check for clashes.

Our category is a string value so not sure we could use it as an additional encode value.

If we start running into clashes we can always prefix the generated hashID with something to scope them.

Appreciate we're using it for ID generation rather than ID obfuscation 👍

[ullmark]

@ckpearson Ok, I'm obviously not informed into your details enough to really know, but from sound of it, you're trying to use Hashids for a task that it might not be best suited for... but with that said;

I would still recommend you to have an integer id stored together with your data that you can use hashids with to encode. That's what the library is for, not generating id:s.