Prisma

Pls, follow guidelines and stick to threads :slightly_smiling_face:

hey, prisma-binding has `fragmentReplacements` option in constructor ( <https://github.com/graphcool/prisma-binding#constructoroptions-prismaoptions-prisma>)  but I can’t find any examples of using it :disappointed: Do you have any examples?

<https://github.com/graphcool/graphql-server-example/blob/master/src/resolvers/Home.ts>

hello, just to chime in with my doubts about the different approaches found around Graphcool/prisma/apollo.

- Not sure if Graphcool is being replaced by prisma or the corresponding Use cases

frankly I am not sure either, but Graphcool has lost it's magic for me, it's way too opinionated and closed causing a lot of headaches to me ... prisma is transparent and well done

I can imagine that Graphcool can help a lot someone entering GraphQL world, it's super easy to start with as long as you don't need custom resolvers and other things. So yea I think that Graphcool should not die completely

- Couldn't find a subscription example working with prisma

what trouble you have with subscriptions, yes there is a little examples, but it turns out it's not that complicated

i miss a client side example. This are only server side tested with the playground. <https://github.com/graphcool/prisma/tree/master/examples/subscriptions>

well, client side is not Prisma specific, depends on what client you are using, if Apollo, look around there :slightly_smiling_face:

- Not sure how to deal with authentication ( if i use graphql for that or some API gateway level auth )

prisma layer has its own authentication (if enabled)
how you implement authentication for your application layer is pretty much up to you ... you can use template (when prisma init) which contains auth to give you some idea

i tend not to use Prisma or GraphQL for authentication, i see benefits in having it bundled with other features like rate limiting and routing different APIs

Anybody coming through the same questions would greatly help with your considerations/examples

Hey, i wanted to dive into prisma but when creating a project via prisma init - and selecting the react biolerplate - the server wont start

<https://github.com/graphql-boilerplates/react-fullstack-graphql/commit/b0d37a0eb884c103d77936303dbf96c4fe64b617>

wich caused npm packets not be get installed properly

good! :slightly_smiling_face: was this communicated differently in the README or instructions?

has some react errors on it preventing it to render

<https://www.prismagraphql.com/docs/reference/prisma-cli/commands-aiteerae6l>

can you describe again what should be updated re `yarn start` vs `npm run`?

Next steps:
  1. Change directory: `cd hello-world/server`
  2. Start local server: `yarn start` (you can now open a Playground at <http://localhost:4000>)

to fix it i ran npm install withing the server dir

thanks a lot! :slightly_smiling_face: I'll try to reproduce this as well

oh also deleteing all workspaces from the console will break that as well

I'm not sure I understand the context here

and deleted the workspace that was there by default

now i cant access the console because a react error stops the app from rendering

please reach out to `support@graph.cool`. Thanks! :slightly_smiling_face:

complaining about qutation marks and such

Hey - so I’m trying Prisma Cloud for the first time right now and i’m finding it’s still really slow

<@U0RQY0KK5> <@U0MQEEY4V> is that normal? Or is there new ratelimiting?

It was supposed to change to 10 requests / 10 seconds.  I'll be interested to hear what they say.

if I wait like 10 minutes and do a request, it’s still super slow :confused:

<https://gyazo.com/025a7764bf0931a39a779d71b97d28f2>

<@U9TSDJV1S> what happens if you run it 3 times straight?

Thanks a lot for bringing this up, <@U6KUJ0RDF>. It seems that the request throttling does not work as communicated for all services. I will look into patching that up soon.

<@U0RQY0KK5> okay so this is why I would have slow response rates yet?

That's correct <@U6KUJ0RDF> :slightly_smiling_face: I'll let you know once this is fixed!

<@U6KUJ0RDF> we rolled a fix to throttling. The 10reqs/10s as communicated is now actually in place.

When writing a mutation resolver for Graphql-yoga -&gt; Prisma, what's the best way to ensure integrity? For instance, I want to move a coin from player1's account, if it exists, to player2's, but not have problems with coins replicating or disappearing.

I think I've read somewhere that nested mutations are atomic

<https://www.prismagraphql.com/docs/reference/prisma-api/concepts-utee3eiquo#transactional-mutations>

haven't tested that tho, not sure how reliable it is :slightly_smiling_face:

A paragraph in the docs is all the assurance I need. :squirrel:

So how would you wrap multiple updates in the same mutation? If you do something like this: ``` mutation{
  a: updatePlayer(
    where:{name:"p1"}
  	data:{totalCoins:2})
  {
    name
  	totalCoins
  }
  b: updatePlayer(
    where:{name:"no-exist"}
    data:{totalCoins:3}
  )
  {
    name
    totalCoins
  }
}``` "a" goes through even if "b" fails.

That's an interesting discussion, thanks for bringing it up <@U9FNEC075>. We are thinking about supporting transactions across multiple top level mutations. There has been some discussion about that here:<https://github.com/graphcool/prisma/issues/74> Would love to hear your take on this in the issue :slightly_smiling_face:

Harmony and I were discussing this in a thread today. One of the problems with this case would be double spending. It looks like you would need to wrap a query as well as both mutations in a transaction to ensure that the resolver wasn't hit multiple times. It may be better have the resolver make a custom sql call directly on the database or through Prisma, if that becomes a feature. Especially if you start doing more complicated exchanges.

That's certainly a possible approach. We're exploring raw SQL queries as a general fallback strategy already, and will detail our thoughts here soon: <https://github.com/graphcool/prisma/issues/2052>.

I commented on issue 74. To summarize: I can almost accomplish this with nested queries, but can't guarantee that double spending won't take place. If there was something in addition to  `where: PlayerWhereUniqueInput!` like an optional `andWhere: [PlayerWhereInput!]` you would be able to check some other conditions. The same would hold true for transactions across multiple top level mutations.

thanks for your input! :slightly_smiling_face: for now, you can also think about locking the mutation in your GraphQL Server layer instead

Currently, the solution I came up with was adding a `lastState: someUUID` property to the player type, that would be changed on every update. That way, the `where: PlayerWhereUniqueInput` would verify the last known state returned from the preceding query and nested mutation transactions would take care of the rest. I'm not sure if this was the type of pattern the Prisma developers had in mind or not.

Has anyone tried using <http://pusher.com|pusher.com> for subscriptions/ws transport?

Wish I had a better answer.  <https://cl.ly/131u240a3b3e>

Obviously there's a way to parse it from an introspection query, but that may be more trouble than it's worth

Copy + Paste it is :slightly_smiling_face:

personally I am using graphql-code-generator to help with that

with Typescript I get a real enums with that generator, no strings, works really great

I don't understand your initial question, <@U6KUJ0RDF>. Can you elaborate?

We're looking for a way to expose an enum list to the client.

This might help: <https://www.graph.cool/forum/t/how-to-access-the-possible-values-of-an-enum-type-created-inside-the-console/23?u=nilan>

The answer is over a year old :slightly_smiling_face: But should point you in the right direction.

I would argue a hardcoded list in your Yoga server might be the better approach.

Is there a way to map a column name to a schema field? 

```
const List = new GraphQLObjectType({
  name: 'List',
  fields: {
    id: {
      type: GraphQLID
    },
    name: {
      type: GraphQLString
    },

    archived: {
      type: GraphQLBoolean,
      column: 'is_archived' // &lt;--- This thing
    },

    items: {
      type: new GraphQLList(Item),
      resolve: listAllListItems
    }
  }
});
```

I don't understand your question. can you elaborate?

How do I know my local target project id? This is an example from docs to add to target:  ```  dev: shared-us-west-2/cj8n7vc0101ya01424ihgryke```

I deleted my existing local .graphcoolrc and ran graphcool deploy. I was able to create a local endpoint.

However when I open the given endpoint  in the browser playground it says: "Schema could not be fetched.
Please check if the endpoint '<http://localhost:60001/simple/v1/cjf8npla30006019252wl4rbi>' is a valid GraphQL Endpoint."

Does the Prisma Cloud UI work for anyone? I can’t update anything

we deployed a fix for this, thanks so much for reporting it!

Screen Shot 2018-03-26 at 4.17.02 PM.png

Maybe your schema is broken? Did you recently deploy a new version of that project to Prisma Cloud?