search

hasura / graphql-engine

Blazing fast, instant realtime GraphQL APIs on your DB with fine grained access control, also trigger webhooks on database events.
https://hasura.io
Apache License 2.0
30.95k stars 2.75k forks source link

RFC: Guide to offline-first and realtime sync with Hasura #4190

Open gautambt opened 4 years ago

gautambt commented 4 years ago

We have a tutorial for building an offline first app with RxDB & Hasura at https://hasura.io/blog/building-an-offline-first-web-app-with-rxdb-hasura/

There is also a design guide (WIP) at: https://hasura.io/blog/p/1c5079be-d19e-4fd4-b700-2a6181f3fc40/

Demos: Todo App: https://rxdb-hasura-demo.netlify.com/ Trello: https://trello-offline-first.netlify.com/

Todo: For our RxDB tutorial, we're looking at adding a guide and best practices around:

  1. Error handling: Patterns to detect sync errors and propagate that to the user
  2. How to handle initial loading of a large amount of data
  3. Showcase conflict resolution with Hasura (one with postgres, and one with actions)
  4. Handle data modeling on RxDB (realtional vs nosql).

There is also https://github.com/pubkey/rxdb/issues/2048 that needs to be fixed.

What else would you like to see from an offline first & realtime sync perspective? Comments appreciated! 🙏

altitudems commented 4 years ago

I think this is great!

The more we can encourage offline-first the better. There is a real opportunity for Hasura to lead here for the GraphQL community.

It would be amazing to have a very barebones example as well without the dependency on RXDB.

Maybe modify the Todo app for other frameworks like Vue, Angular.

I know there was a mention of some example apps being created. Would be great to some an offline-first ones.

dhmacs commented 4 years ago

Just an idea, but I think It would be super cool if you build a data store (just like Amplify Datastore) that works out of the box with Hasura. The ideal DX would be like, I take care of changing data locally according to my business logic, and Hasura Datastore takes care of replication and conflict resolution

gautambt commented 4 years ago

There is now a guide focusing on conflict resolution over here: https://hasura.io/blog/couchdb-style-conflict-resolution-rxdb-hasura/

dhmacs commented 4 years ago

@gautambt 404 link

weyert commented 4 years ago

I think the correct url is: https://hasura.io/blog/couchdb-style-conflict-resolution-rxdb-hasura/

gautambt commented 4 years ago

Oops! I have updated the comment above with the right link! Thanks, @weyert @macs91

RodolfoSilva commented 4 years ago

Should have an example with flutter or native Android, maybe a generic way to do that.

barbalex commented 4 years ago

I am in the last days of developing an offline-first app based on the ideas in https://hasura.io/blog/couchdb-style-conflict-resolution-rxdb-hasura. I used to work with CouchDB itself before moving to PostgreSQL because of it's relations, role-management and general greatness. Before reading the article I hadn't realized you could solve conflicts CouchDB-style with PostgreSQL. Thanks a lot for that article.

The app is here: https://github.com/barbalex/vermehrung

Instead of RxDB it uses Mobx-State-Tree which is an awesome tool for the task. Together with mst-gql. I love this combination. You should try it too 😇.

Instead of resolving conflicts on server and client it only resolves them on the server. The client reads winners and writes revisions. This solves the problem of the client having to have all revisions - blowing up storage space and network traffic.

The server adds a list of conflicts to the winning revision. The client then loads those revisions when the user resolves the conflict.

Would love to see a guide.

Aditya94A commented 4 years ago

I would love to see a focus on the devops side as well, managing dev/staging/prod environments, migrations, docker-compose-ing, best practices etc.