Rails in a week - day 1


TL;DR: I began learning Rails this morning, and even though Rails in itself is (seems?) easy enough, setting everything up and deploying is hairier.

8:00: let's get started! First step: getting vagrant up and running. We'll hit the tutorial. 8:15: the lucid32 “box”, Vagrant's parlance for a virtual machine image, is downloading. Time to get a cup of coffee. 8:40: box downloaded, let's get on with the VM setup process. vagrant ssh… Yep, it works! 8:50: adding a few cookbooks. The Vagrantfile syntax (which is actually Ruby) isn't recognized by vim; to fix that later. 9:00: first oops of the day:

[default] Failed to connect to VM! Failed to connect to VM via SSH. Please verify the VM successfully booted by looking at the VirtualBox GUI. [09:06:25] florent@Air:~ $

The VirtualBox GUI shows the VM running, but nothing more. Can't force restarting the VM (or even stop it) from the GUI. Let's start over: kill the VM process, vagrant destroy; vagrant up

Same error. Uh-oh. Is it because of the cookbooks I added? Let's try deleting them and reverting to the original Vagrantfile. No luck.

Alright! Google to the rescue! And there we go, a Stack Overflow discussion leading to a bug report on GitHub. Well known network-related issue then, a fix seems to start the machine with GUI enabled and /etc/init.d/networking restart so that vagrant can SSH into the VM. Not ideal, but meh. Let's advance! 9:45: phew! That took some time. Now let's add back those cookbooks. 9:55: added to the Vagrantfile, the cookbooks install themselves. 10:00: port forwarding works, now's time to go buy groceries while downloading a Debian Box for later.

13:45: back! Stomach full and coffee by my side. 14:30: that whole Vagrant/Chef thing is a bit strange. There seem to be a recipe for Rails on Opscode's (the company behind Chef) GitHub account, but it seems to also installs a bunch of Java stuff. Anyway… We'll get to the bottom of provisioning later, the goals here is to learn Rails, right? Let's just install what's needed by hand. 15:00: new Debian Vagrant box set up. Installing rvm to get ruby 1.9.2. 15:20: rvm install 1.9.2 then rvm use 1.9.2: Ruby all set. Good. 15:22: gem install rails let's go!

(Starting Rails for Zombies on the side) Lesson 1: okay, so there's a built-in ORM in Rails. That seems to be ActiveRecord if I understand correctly. Gives you methods like Tweets.find(id), etc. Lesson 2: models. Models are the O in ORM, and the M in MVC. Lesson 3: erb. Built-in templating. The V in MVC. Lesson 4: controllers. The C in MVC. Lesson 5: routes. The mapping between URLs and actual code.

17:00: alright! Rails for Zombies is done, I feel ready to start a real Rails project. rails new sample Bunch of stuff getting created… All done. Let's launch! cd sample && scripts/rails server … Crashes. Says it needs a Javascript runtime. Why? No idea. But here's the fix. “Still pretty lame that rails 3.1 is “broken” out of the box.", says wonderfulthunk. Quite true. :|

17:20: gem added, bundler works (bundle install manages dependencies and puts the needed gem into the “/vendor” folder), script/rails server works! Let's plug the host's port 3000 to the VM's one and set up a shared folder so that we can develop and test from the host machine while running everything in the VM. This is all done in the Vagrantfile. 17:30: setting up the shared folder ate the Rails project. Fun times. Re-create it, re-add the gem, re-bundle, re-start the server… 17:40: it works! I can see Rails’ welcome page. 17:41: so, what now? “1. Use rails generate to create your models and controllers” Okay, sure. Let's try and create something simple, say a blog. It need posts. script/rails generate model post: it creates the model I want, some migration (?) stuff and some testing stuff. But wait, it doesn't create any view, or controller… There is a better way: scaffold. script/rails generate scaffold gives us an example of how the command works, by suggesting to create… A blog post. Great minds think alike, I guess. ;D [/narcissism] Scaffolding creates another bunch of stuff. Looking at config/routes.rb, there is now a resources :posts. So I guess going to http://localhost:3000/posts should work? => Could not find table ‘posts’. Oh. Let's see what's in the db then. I remember an option on the rails script about that: yep, script/rails db It gives an SQLite shell:

Ah! So there's no database. Time to learn a bit more about that migration stuff. 18:15: So I checked the official Getting Started guide on RubyonRails.org, recommended by orta on HN, and it's really well written and comprehensive. I should have started here actually; it's exactly the right amount of conciseness and straightforwardness. That'll teach me not listening to others. The database is created with rake db:migrate. 18:20: http://localhost:3000/posts is now a fully functioning CRUD app. Is it supposed to be that easy? It really feels like cheating. The example on the aforementioned Getting Started guide is a blog, so I'll piggy-back on it for the rest of the day. 20:00: hmm. Learned about automatic code generation, configuration over convention (it's all automatic! The error messages look like they might be a bit cryptic from time to time, though), migrations, and partials. Lots of nifty features indeed. And all is quite simple. So simple that I'm actually going to try and quickly write that antipodes application tomorrow, then practice TDD and stop worrying about deploying before I actually have an app to deploy.

On a side note, yesterday's “day 0” post has been viewed nearly 1200 times thanks to a (brief) appearance on HN's front page. That helps building up some pressure, I hope not to disappoint. :)

Come back tomorrow for more bug-fighting, stumbling in the dark and unstructured write-ups!