Developer Preview: Graphcool 1.0 Beta


Hey @ddub, thanks so much for your feedback!

It’s true that now you have to get familiar with more concepts than before. However, this also gives you access to a vast number of new capabilities and at the end of the day reduces the effort you have to invest to accomplish the same :slight_smile:

Furthermore, the GraphQL Boilerplates do get you up and running quickly so you can dive into developing your application. Let me know if you have any feedback for the boilerplate experience; I’d love to incorporate it!

– Nilan


Hey Steve, sorry about the confusion! :slight_smile: Let’s clear that up!

Graphcool Cloud will continue to exist; you will be able to deploy services to a free development cluster, or to a cluster on platforms like Digital Ocean, AWS or Azure.

The migration being talked about in this thread is migrating a Framework service to a Graphcool 1.0 service. We’ll release more material about that soon. Furthermore, we will continue to support the Graphcool Cloud for the Framework, too :slight_smile:

Do note that the free development cluster runs on shared resources. For larger apps we recommend hosting Graphcool on a dedicated cluster instead.

Regarding your second post - what you can do with resolver and operationX functions in the Framework is much simpler accomplished by a plain GraphQL resolver in Graphcool 1.0.

Instead of relying on a strongly opinionated developer experience and API when working with Graphcool Functions, you can now rely on the best practices in the GraphQL Ecosystem.

Note that the concept of subscriptions triggering a webhook continues to be present in Graphcool 1.0 :slight_smile:

I do hope that my answers were helpful, let me know if you have further questions!

– Nilan


Hey James, we’re continuously improving examples and documentation! Feel free to provide your feedback in the form of issues or PRs to the documentation or examples here.

Thanks :pray:
– Nilan


Thanks, I updated all links in the original post :slight_smile:


First of all “Loved the approach of graphcool-db and graphql-binding”, its an awesome combination of giving us the power and flexibility, balance is just right.

What I need help now is to understand how do I run this in production on docker? So here are my questions

  1. graphcool command line tool is great for development, but I won’t be using it for production on my docker cluster
  2. I tried “$>graphcool local eject”, which I loved since now I got my docker-compose.yml
  3. My confusion is about “How do I do $>graphcool deploy after I am running things on docker-compose.yml”?
  4. Does the graphcool-db (docker based microservice) provide any apis for me to upload a new graphql schema and control how the migration plays out?



Are subscriptions working yet? Or should I wait?


I have tried subscriptions and they are working

Add following in src/schema.graphql

    type Subscription {
        post(where: PostSubscriptionWhereInput): PostSubscriptionPayload

And following in index.ts (for typescript example)

    Subscription: {
      post: {
        subscribe(parent, args, ctx: Context, info) {
           return{}, info);


Is any of the “subscription/relationship edge” updates making it into the new drop with Framework 1.0, as discussed here @nilan:


Any updates on the how to do graphcool deploy with ejected docker-compose.yml?


Yes, when developing features for some clients certain trade-offs need to be made. I think in this change to 1.0 a lot of features that worked great for some users were forgotten about as new features were implemented for other users. It’s your guys business so obviously build this framework for who you want. But a lot of these new features are less important to others, while ease of use may be more important to some.

Keep in mind I also have no idea what your plans are for a cloud version, so maybe I’m speaking out of turn.

But overall what I’m trying to say is, I probably wouldn’t be using GraphCool if I had originally come to it for the 1.0 version. As the quick on-boarding and ease of use was what got me hooked in the first place.

Even with the boilerplates, it took me a couple hours to get up and runing. Partly due to a bug and confusion from the documentation. But with the original hosted version, I was up and running in minutes and because it’s a hosted app I wasn’t bogged down by environment inconsistencies and didn’t have to read too much documentation to get going. I think that’s a really big deal!


On one hand, I agree - while I haven’t used the original Graphcool, it wasn’t zero to 60 after a few CLI commands and a few pages of docs like I experienced with Vue. That said, I think that is less a fault in the new framework itself, and more current fluctuation due to active development, ongoing work on bugs/ergonomics, and a need for clearer/more developed/more consistent documentation.

The staff can overrule me here, but I’d imagine they’d be more than happy to get feedback on specific issues in onboarding/documentation/APIs/etc that could be addressed. If no one makes suggestions for improvement, nothing gets changed! I’m planning on doing that myself as I have the capacity.


Part of the quick on-boarding experience of the original Graphcool was the fact it was a SaaS project. So I could start off with out having to set up or manage any server instances or databases myself. And the GUI that came with it made it very intuitive to start creating models and set up roles as a new user.

Maybe the future will bring this type of system back, but for now it’s gone from Graphcool.

On the home page of there is a quote.

Graphcool is a really easy way for front-end devs to get started with GraphQL

— Lee Byron, Co-creator of GraphQL

I don’t know if this holds true anymore.


You should still be able to use Graphcool as SasS - I just don’t know/remember what the current status is of their service during the beta (are all deploys to cloud clusters completely public until authentication is finished? Or was that taken care of or a misunderstanding?).

As far as the GUI, I definitely see what you mean now that I’m trying it. It looks a lot like the playground, but certainly carries more advanced features as far as being able to define almost everything through the UI. Maybe this is purely for their Saas product? If so, once 1.0 releases I imagine it will be available just like 0.x. Otherwise, hopefully it’s just a matter of time that it gets bundled into the playground.


Ya right now GraphCool has 3 distinct products in limbo. It’s not easy for the casual user to keep track of what does what right now.


You need to on it from their side. Do you think is easy to work with old version in bigger teams? How many small projects they need get running to get money from one bigger company. They are still trying to make living from it. I started from GUI version too, but was kind of skeptical with framework and then with 1.0, but now i love it. Why? Because you should be programmer and write code and have everything under control not clicking on GUI, what is kind of slow to do stuff and work with more complex tasks.


Totally agree with you @Huvik. Having all of this under source control ultimately means you can move much much faster, with much higher degrees of safety.

It’s all about where you want to pay the price of time. Up front with a simple GUI, or a much bigger price over the long run constantly trying to keep some UI based system in sync with your product.


However, as a note:


For anyone still referring to this thread - Graphcool 1.0 Beta has been renamed to “Prisma”

See release post:


Thanks everyone for providing your great input. We have released Graphcool 1.0 under the name Prisma earlier this week.

If you have any questions at all about the Graphcool Framework and Prisma, refer to this thread: Graphcool Framework and Prisma

Steps to upgrade to 1.0