I have yet to be approved (I applied a week ago) but this looks very promising, even though it could end up being costly - not having to manage servers, GraphQL, and scalable real-time data are all bug wins. No one talks enough about GraphQL subscriptions IMO, I posted about it earlier but not enough interest.
From what I’ve seen and read here Meteor’s default mode of pubsub for everything ends up being a liability in many cases.
I do have the same feeling about what other people feel about realtime pubsub, but people highly underestimate the improved user experience when done right.
In my case i have projects that are event sourced, built with the CQRS approach or a flavor of them. They require a mechanism to notify users when a state jas updated. Meteor and pubsub are a major asset in that.
Its understandable that most people don’t want to use pubsub for standard websites. Its just overkill. Then again I find even React or Vue to be overkill when it comes to standard websites. I’ve build the public site for my saas platform using a simple bootstrap template and some added html and custom styles within a day and am hosting it in an AWS bucket…
In the end its about the choice of tools for the job to be done. Building a saas? Appsync might be a major asset
Looks like a good example of how a combo of Meteor’s livequery and GraphQL could look like. Hopefully MDG can provide something similar in the future for Meteor.
What do you mean? Seems like because Apollo is on Amazon as AppSync now, that we can now make Meteor apps with GraphQL backends on AWS? Or is that not possible yet?
Yes, but then you’re locked in to amazon’s proprietary appsync and have to store/query all your data with appsync.
In that aspect it is similar to using firebase. I meant for Meteor to support easy graphql realtime updates in a self-hosted / open-source solution.
You could ofcourse ask what locked means. If it gives you all of this with zero dev time, then you will not be locked at all. You simply follow the graphql specs and migrate as soon as you want to host your own solution.
This is exactly what I am looking for. I guess the missing part is MiniMongo to store and merge data on the client side? Or is there other obstacles that prevent this? I am asking because I am pretty undecided if I should still go for Meteor/DDP/MiniMongo or try Apollo subscriptions instead.
Im no expert, but I believe the caching layer in Apollo kinda covers this. The biggest hurdle to me right now seems to be about optimistic updates with mutations.