Adam Loving adamloving

Hosting Node.js

We need to host node ourselves (Rackspace server) to have access to the file system (temp files for image processing) and (legacy) local database.

Have you hosted node in production? If so, what was the web server setup? Currently, we're using nginx + a unix socket file + forever or upstart. My inclination is to stop hosting it ourselves and go for a PaaS (like heroku) - but that's because I wish this stuff was not my problem. What I'm looking for is a clear recipe for best practices.

Issues

Hosting node is hard because unhandled errors force the node.js v8 process to die and need a restart.

Guidelines

Only nest the simplest of functions
Give functions a name
Seems weird to pass callback function as an argument

The callback way

async.parallel([

When you've got a pile of bugs and features, it's usefull to separate severity from priority in order to prioritize them. At the end of the day, priority is all that matters - but severity helps clarify how "bad" the bug is.

Basically, it comes down to...

SnP1 = hotfix (where n is any number)
SnP2 = in this sprint
SnP3 = later (maybe never)

... but almost always, high severity bugs will have a high priority.

Where is apache?

$ which apachectl

It's not a symlink, so it must be mac version! Where's the confthing?

$ apachectl -V | grep conf

Where's PHP?

Rails was designed to provide an opinionated way to solve common web application problems. Node has many of the same capabilities, and a couple frameworks have sprung up to apply the same patterns. Here's what I've missed from rails that I hope to find in a Node framework.

agreed (“opinionated”) file structure
conventions for file names
data migrations (code files that make db schema changes)
built in auto-compile for Coffeescript SASS/Stylus/LESS and HAML (no need to grunt)
asset pipeline - versioning of JS and CSS resources, spriting images
ORM (ActiveRecord): same/similar semantics for MySQL, Postgres, or MongoDB
Capistrano (or other scripted deploy)

	#!/usr/bin/env ruby
	# foreman run script/social/auto-tweet.rb data/crawl-data.json
	require 'Twitter'
	require 'json'

	client = Twitter::REST::Client.new do \|config\|
	config.consumer_key = ENV['TWITTER_API_KEY']
	config.consumer_secret = ENV['TWITTER_API_SECRET']
	config.access_token = ENV['TWITTER_ACCESS_TOKEN']
	config.access_token_secret = ENV['TWITTER_ACCESS_TOKEN_SECRET']

	#!/usr/bin/env ruby
	# foreman run script/social/auto-fbpost.rb data/crawl-data.json
	# To get page access token, go to graph explorer, get user token with manage_pages and publish_actions permission,
	# then request /{userId}/accounts
	require 'koala'
	require 'json'

	graph = Koala::Facebook::API.new(ENV['FB_PAGE_ACCESS_TOKEN'])
	urls = JSON.parse(File.open(ARGV.first).read)

-function batchDelete() {
-  var batchSize = 10; // Process up to 10 threads at once
-  for (i = 175; i > 0; i--) {
-    var threads = GmailApp.search('label:outbound older_than:' + i + 'd');
-    for (j = 0; j < threads.length; j += batchSize) {
-      GmailApp.moveThreadsToTrash(threads.slice(j, j+batchSize));
-      // Logger.log('finished move %d', j);
-    }
-  }
-}