I have a business case where I need to have documents in a collection containing a number field that will be unique than another document's number field. This number has to be short to type and should be random. For example: 2832. Now I know this doesn't scale since there can only be 10k possible number combinations for a 4-digit number, however, that's fine at the moment.

I wanted to know the best way to implement this in Firebase. I was thinking of creating a 4-digit random number in JavaScript then whenever a new number is generated, do a check on a separate collection that contains one document with an array field that will contain every unique number. If that number already exists, generate a new number and check again, if it doesn't exist, then add that unique number to the array.

Is this the optimal approach? Thanks.

CodePudding user response：

Whether something is optimal is hard to say, but the approach sounds like a reasonable starting point.

I'd probably have started with having a separate document with the number as its document ID, but for 10K values that might not even be needed - which means your approach will be simpler.

Be sure to use a transaction to perform updates to the document, and use security rules to ensure no client can write a value that's already in the array.

CodePudding user response：

Well, since you are already open to maintaining an array of used numbers, ou could eliminate the "retry" factor by using the full pre-randomized array and a "pointer" which is a simple number from 0-9999.

Firestore does manage data contention : https://firebase.google.com/docs/firestore/transaction-data-contention

So getting the current pointer value and incrementing it in a transaction would make it so the pointer value is used once only and allow you to get a guaranteed unique value from your pre-randomized array at that pointer value position.