Session overview

The meteor way of sending data
Minimongo or how we have 'MongoDB' on client
How it all works
Customizations
Advanced uses
How publications work under the hood

Normal WEBSITE (json api)

Client (browser) starts request
Server handles the requeest and sends data formatted in JSON
Client library inserts the data in the page

Meteor

At load meteor start syncing data with the client
The data is already on the client when it needs to be displayed (prefetch)
The client can render the data without the need of a round-trip to the server
Operations can happen both on client and server (latency compensation)

Minimongo

Minimongo is a JS library that emulates nearly all the possible operations you can do on collections (only some geospatials query are excluded).
The raw data are transferred from the server (real MongoDB) to the client via DDP (Distributed Data Protocol) using a WebSocket
Minimongo cursors are reactive data sources
You can work seamlessly and simultaneously on both Client Collections (Minimongo) and Server Collections (MongoDB)

client-server architecture

Server

Client

__________________________

DDP

MongoDB

Meteor.publish()

Minimongo

Meteor.subscribe()

How it works

Example 1

A basic publication and subscription

Publications are the way the server exposes the Data that the client may need
Subsciptions are how the client subscribes to the publications and let Meteor sync the data between the client and server.

things to note

Subscription reactivity, polling vs oplog
Observers and their reuse

Customizations

Example 2

Send only the data the client needs.

Example 3

Send only the object fields that the client needs

BEWARE

Multiple publications over the same collections may lead to unexpected results because the data are merged in the client.

This is because of the mergebox.

Advanced Uses

Example 4

Hide some data to non registered users.

It's possible to make publications smarter and differentiate the data that are sent to the client.

BEWARE

You should try to avoid having a publication with multiple possible values in queries.
Meteor does reuse the observer for successive requests, having different query make this optimization impossible and may lead to slowdowns.

Under the Hood

Example 5

How a publication works internally and some nice tricks that you can easily accomplish.

Bonus point

All the customizations do not change the query, so you can actually do this without having an impact on the performance.

So Long, and Thanks for All the Fish

The code used in the presentation is available at
https://github.com/matteosaporiti/Meetup_19_03

Slides will be available later on.