Choice of database about editor-backend HOT 8 CLOSED

We went for using GitHub gist as the database. 😅

from editor-backend.

@domoritz @siddhant1 I tried to research as much as possible but I would really like your suggestions on this.

from editor-backend.

Thank you for writing this up!

https://github.com/elastic/elasticsearch-js is an up-to-date library for Elasticsearch. I also found that you can search over specific fields.

One other issue we need to look into is ranking search results. As far as I can see it, Mongo or Postgres don't have a good way to rank search results. Github and Wikipedia are using Elasticsearch for searching. I see that Mongo or PG are a better fit for data storage but I'm not convinced that they make a better search index.

from editor-backend.

@domoritz Yeah, their usage seems pretty similar to MongoDB only. But storing user information can only be supported in PG or MongoDB. bl.ocks.org were also using MongoDB for authentication and elasticsearch for searching.
So I am still unsure how searching will happen when things are stored across two collections in elasticsearch DB(here users and gist).
Blocks was able to do it since it had the user id as the GitHub handle itself so it simply requested GitHub api for its gist. But in our case, I am planning to have a custom user_id which will be unique. This will ensure extending this platform in a way that the code of visualizations can also be stored somewhere else than just GitHub gist.
Having a user model will help us extend the platform. Otherwise, I guess we can just have one document for gist which will store the gist information and the user it is uploaded by. The latter case will require the user to be registered from one single platform only.

from editor-backend.

I see. How about we go the simple route for now and only support github? Then we don't need to store the user and we don't need to do our own authentication. We hopefully can spin up a service in a month or so and then take it from there?

from editor-backend.

@domoritz okay ✌🏽 I will research a little about the ouath process as well. I think we would only need to create session variables on front-end when authenticated. Need not store anything more.

from editor-backend.

Instead of writing the whole authentication process ourselves , have a look at passport.js

from editor-backend.

@siddhant1 yeah, obviously. I was talking about the response given back by GitHub API after authentication. Will we need to store any relevant information from it?

from editor-backend.