brenapp / vexdb Goto Github PK

This library is designed to be isomorphic and in order to do that, I have used axios, which is an isomorphic Promise-based fetch library. However, it is 10Kb big built and gzipped and has many features which are not required for this application. I will look to create a tiny promise-based library to replace it

Steps to reproduce:
Insert a debug line into the request function to see what network requests are made:

async function request(endpoint, params) {
    if (cache.has(endpoint, params)) return cache.resolve(endpoint, params);
    console.log("GET", endpoint, params);

const vexdb = require("vexdb")
vexdb.get("matches", {
    sku: "RE-VRC-17-3805",
    team: async team =>
        (await vexdb.size("teams", {
            team,
            region: "South Carolina"
        })) > 0
}).then(console.log)

Run the code and see that the request for GET teams { region: 'South Carolina', nodata: true } is being called for every match evaluated

.getAll() will throw an error if no parameters are passed

  1 failed

  getAll » .getAll() resolves for correct requests
  C:\Users\bmcgui7575\Documents\vexdb\main.js:185


   185:             (v, i) => get(endpoint, Object.assign(params, { limit_start: 5000 * i }))


  Error: Cannot convert undefined or null to object

  Promise.all.map (main.js:185:44)
  size.then.items (main.js:184:12)

This is because of the lack of || {} after params

The current iteration method is clunky, and could be optimizied

I've resolved the problem down to the permutationsObject() function. I think it's the result of my array spread

It's pulling down Code Coverage! ;-)

Right now, size is really stupid:

async function size(endpoint, params) {
    return get(endpoint, params)
        .then(res => res.length)
}

This can be improved by:

• Sorting parameters so more nodata: true requests can be made
• Making more raw requests

vexdb tries to do too much for the user. vexbot, which is one of the most in-depth uses for vexdb never uses any of the advanced features, and basically only needs caching, nice type support, and getting all records on a request. Right now, vexdb is trying to do too many things, and it's inflating the source code.