Prisma

<https://github.com/prisma/prisma/issues/3888>

Hi Om working on an app to keep track on chores that we have to do in our household and would like some critique on my datamodel. The chores can either be in "upnext" or "archive" optionally they can also be in "todo" - "done". That is a chore can be
1. upnext or archive + done
2. upnext or archive + todo
3. upnext or archive

-.js

I could alternativly make an ENUM state on each chore I guess - they will be fetched by this state often

midlware_graphql.js

Hello friends
i want to add midlware to graphql resolve
whats the best strategie

Is it possible to have a Type resolver that instead of resolving single fields, but the entire object? My use case, I'm storing an ISO country code in the postgres database, but the graphql data actually comes from static JSON data.

When I use `prisma.createPost` it is returning all of the Posts scalar fields but not things like Author, is this normal?

If I have a legitimate schema type called "Subscription" (that holds user's paid subscriptions), is that going to mess up my ability to run Prisma/graph subscriptions?

(I'm thinking it does, just took me a while to get to this…!)

it may get confusing to read buti believe it will be fine

Is it possible to import a type from a prisma generated graphql file and then modify a field of it?

From what I've been doing - I've had to copy the entire type into my own typeDefs and then make the modification.

I was hoping I'd be able to maybe inherit and then override

We know about this. You can have a look into the future here that we will solve this: <https://github.com/prisma/nexus-prisma>

I have a question if this is the right place to ask?

Anybody know if there’s a way to protect the Prisma client by specifying which operations to generate? For instance you don’t want to ever allow users to be deleted from a client that’s accessing Prisma?

That’s currently not possible. What’s the reason you want to hide it?

The idea would be that from a security perspective you don’t want to expose all of your database operations to every client. 

I dont think you're supposed to expose prisma to any clients :panda_face: only your backend

when I say client I mean AppServer or a lambda, not frontend

The app server is under your control and should not be accessible to users other than through a well defined API. If they can hack your server and make those kind of changes then you got bigger issues on your hands

Is it possible to add more “where” searches than just ID?

Any unique value can be used <@UFGRJ6WRY> 

ok thanks a lot, just not specifying @unique on them currently. thanks!

Quick question, if you’re running Prisma w. Yoga, is there any reason to choose MySQL over MongoDB and vice versa?

It depends on what you want to build. The main reason to use MongoDB over SQL are embedded types. If you think those are important for you go with Mongo otherwise use SQL.

Or if you want total order isolation (required by some laws) then pick mysql/postgres. I use postgres for all my prisma stuff, it has been smooth sailing.

greetings, would anyone possibly be able to help me create a really simple subscription resolver that sends a message every second? My attempt so far is <https://pastebin.com/wutBqskv> but i'm brand new to iterables in js :disappointed:

This might help: <https://github.com/tc39/proposal-async-iteration#the-async-iteration-statement-for-await-of>

Also, MDN is a great source to understand how things work. There are always examples. :wink: <https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/for-await...of>

How do we disable "subscriptions" on the server?

I'm a Prisma noob, so don't know the answer. But why would you want to disable subscriptions?

Usually you would have another GraphQL server infront of Prisma and that would dictate what would and not would be accessible.

Subscriptions add additional schema to the schema file when imported into graphql tools such as makeExecutableSchema. If you dont use the subscriptions, you can see an order of magnitude in performance when you server loads that file. Depending on your scenario, this may or may not be an advantage to you. For our team, running several serverless functions that load that schema for highspeed processing, that additional load time can be a huge burden.

We have done tests with our particular schema, and removing the subscriptions by hand, our schema loads 4.5x quicker.

Woah, that is a very good point and a lot of wasted resources. Maybe make an issue about this on Github with your finding? Would be nice to be able to configure what features are needed. – Sorry I don't know anything about solving it, except run post processing on the generated file like `gulp` or similar tools to remove what is not needed.

wow thats an ugly new slack logo... anyways is anyone storing  comment threads in prisma?

I am curious how you stored it schemawise

I will soon be diving into setting up a similar schema. If you find a good method it would be nice to hear.

I guess I am just going to be doing something like

```
Comment {
    parentComment: commentID,
    commentText: String,
}
```

What about doing:
```
Comment {
    parentComment: Comment!,
    commentText: String,
}
```
Then the relationship is clear, and if a comment is deleted the children can automatically be cascade deleted as well.

yea, what you wrote is actually what I meant to write

it should definitely be a relationship, not stick the id in there

I just wrote it by hand into the slack window haha

Cool to hear back if you encounter any challenges with this approach. I will probably be testing it out myself in a few weeks as my project progresses.

sure ill let you know, I doubt I will see issues initially, my concern is scaling as comment threads get large with a lot of reads

Then you should have two distinct types of comments. I.e: a `Comment` can have a field `replies: [Reply]` but replies do not have anything else but the content of the reply itself

That will ensure that your threads will only grow linearly in comments

is it possible to have PK with UUID instead of ID?

Hi! I decided to try using Prisma for the first time and  encountered a problem regarding adding business logic to resolvers. Suppose I have a mutation `createOrder`. A user can only create order if they have less than 5 active orders

I think you misunderstood the purpose of Prisma Data Access Layer (DAL).

After you design your data models, it's exposing a graph that enables u to query/mutate the data with his fluent API or directly through GraphQL
Decoupling the underlying database related complexity/logic (SQL, NoSQL, syntax, aggregations, etc).

Instead "adding business logic" is not in the scope of the DAL, you must handle this need in different ways:

- on Client Apps:

client apps (business-logics) =&gt; Prisma =&gt; Database

you can directly call Prisma from your client apps that are in charge to handle the business-logics:
before creating the order, you've to count the number of already placed orders and decide if the user can or can't place the order.

- on Application Server

client apps =&gt; <https://myapplicationserver.io/> (business-logics) =&gt; Prisma =&gt; Database

you need to create another service in order to expose a REST-API or a Graph with your business logic.

```

const resolvers = {
    Mutation: {
      createOrder: async (_root, args, ctx) =&gt; {
        const { prisma } = ctx
        const { order } = args

          const ordersCountPerUser = await prisma
            .ordersConnection({ where: { user: order.userId } })
            .aggregate()
            .count()

          if (ordersCountPerUser &gt;= 5) {
            // handle your fail case
            throw new Error('Can't place more than 5 orders!')
          }

          return prisma.createOrder(order)
      }
  }
}
```

some examples HowTo:

- <https://github.com/prisma/prisma-examples/tree/master/node/graphql-schema-delegation>
- <https://github.com/prisma/prisma-examples/tree/master/node/rest-express>

So, how to count order amount for a user inside a resolver function?