I have a general question that might be more of a SQL question. I want to create a row in my Transaction table, but my 3rd party API asks for the ID of the Transaction before they create an entry, but I want to do it in parallel. So my general solution would be to create a Transaction, send it to the third party API, wait for a response or error, delete the row with the ID if there is an error, or correlate it if it's successful. Is there a way I can just create the ID for the Transaction table in Prisma and conditionally discard it if there is an error with built in solutions?

Something like...

goToApi: async (_, args, ctx, info) => {
   const transaction = await ctx.db.mutation.createTransaction(data: {... })

   const api = await fetch(`third party`)

   if (!validated api) ctx.db.mutation.deleteTransaction(where: { id: transaction.id })

   else ctx.db.mutation.updateTransaction(where: {id: transaction.id}, data: {... api.data}})
}

Hehe seems like what I was going to do. Wanted to see if there was some way of just creating an ID without the whole row 🙂

If you want to be really efficient, you can use a UUID. Are you familiar with that?

@lawjolla I will definitely look at that now 🙂

I can explain it quickly..

If you have time. I can always read the manual.

A UUID is a unique, randomly generated identifier. You can ask the server to generate one. Then you make a unique field with the UUID on the type, e.g.

type Transaction {
   ...
   uuid: ID! @unique
}

Basically, it's a server generated unique ID. So you can later query by its actual ID or its UUID. This is a good pattern for keeping track of things on the client without getting the server id back. The client generates a uuid and can then query the database with that uuid instead of getting the server id

I've used @unique, but I didn't think it was 128-bit?

"A sample of 3.26*10¹⁶ UUIDs has a 99.99% chance of not having any duplicates." https://towardsdatascience.com/are-uuids-really-unique-57eb80fc2a87

I do a lot of crypto development so It's up my alley 🙂

In other words, you can safely assume that a UUID will not have a unique collision, though it's theoretically possible

Have you heard of the birthday Paradox?

If the concern is UUID collisions, mathematicians a lot smarter than us came up with it and the pattern is used all over computer science for nearly a decade.

I'm going to try that implementation you just posted that's super interesting