Nicolás Hock Isaza

Tower of Hanoi with a stack

2013-02-05T14:37:00-05:00

I've recently subscribed to Coding for Interviews and received my first email today. The content is great and you should join. Even if you're not looking for a job, it's a short read with nice topics to refresh your memory.

Today's assignment was with stacks. I've used stacks to reverse a list in O(n) but today's challenge was to solve the Tower of Hanoi problem without recursion.

Here's my solution:

Hanoi = Struct.new(:disk, :source, :dest, :spare) def move(hanoi) "Move #{hanoi.disk.abs} from #{hanoi.source} to #{hanoi.dest}" end class Stack def initialize @data = [] end def push(hanoi) @data.unshift(hanoi) end def pop @data.shift end def peek @data.first end def empty? @data.empty? end end # Resolve the Tower of Hanoi problem with a Stack. You can't use recursion # in it's traditional way, but adding everything to a stack works the same. # # Notice you have to invert the order of the calls to the recursive function # since the stack is **Last In - First Out**. def stacker hanoi stack = Stack.new stack.push(hanoi) until stack.empty? do hanoi = stack.pop if hanoi.disk <= 1 puts move(hanoi) else left = Hanoi.new(hanoi.disk - 1, hanoi.source, hanoi.spare, hanoi.dest) right = Hanoi.new(hanoi.disk - 1, hanoi.spare, hanoi.dest, hanoi.source) hanoi.disk = -hanoi.disk # force printing it next time it appears stack.push right stack.push hanoi stack.push left end end end # Resolve the Tower of Hanoi problem with recursion. def recursive hanoi if hanoi.disk == 1 puts move(hanoi) else left = Hanoi.new(hanoi.disk - 1, hanoi.source, hanoi.spare, hanoi.dest) right = Hanoi.new(hanoi.disk - 1, hanoi.spare, hanoi.dest, hanoi.source) recursive(left) puts move(hanoi) recursive(right) end end h = Hanoi.new(3, 'A', 'B', 'C') recursive(h) puts "\n\n" stacker(h)

There's another iterative solution but I haven't implemented it. So, if you did or have some other way to solve this, I'd love to know on Twitter @nhocki

Stop using config files for secret data

2013-01-25T11:02:00-05:00

Seriously, it's your fault that Github exposed your keys. If you want to add everything to your repo (not that this is a good idea though), at least make it private.

If you want to have per-environment variables, use the environment variables. But my development env would be full of crap! some of you may cry… well, there's a simple solution for that. Use Dotenv.

With Dotenv you only need to add a .env file to your application root directory and you're done. Here's a quick example:

# .env
S3_BUCKET=some_development_bucket
S3_SECRET=a_secret_key

Then in your code, simply use ENV[] and be happy (and safe, but mostly happy)

# some config file
config.fog_directory  = ENV['S3_BUCKET']

And just like that, you can setup your server's config environments and not worry about supposed-to-be-private public information.

One really cool thing about this is that this variables will only be loaded in that specific application. It won't change your environment. So when you stop the servers, your environment won't have the .env variables.

You can read more about Dotenv here.

Easily add versions to your Gemfile

2013-01-15T22:31:00-05:00

Since bundler came out, managing dependencies on ruby applications has been amazing. Combined with some bootstrap script, it's really easy to download some source code, install, and run an application.

There is still some room for improvement though. I've always supported the use of SemVer and pessimistic locking (the ~>). This will allow you to update applications without pain. And, with all this rails vulnerabilities, you need just that.

But, since people don't always have the versions on their Gemfile, this can get a little messy. So, I've created a simple gem to help you with that.

Welcome version_gemfile

1. Simply install it: $ gem install version_gemfile
2. Go to your rails application: $ cd ~/my/ruby/application
3. And add the versions to your Gemfile: $ version_gemfile

How does it work?

This gem will go through each line in your Gemfile looking for dependencies that have no versions. Then, will look into your Gemfile.lock for the version you are currently using and add a pessimistic lock on that version.

The code can be found on Github. If you have any problems, open an issue there or ping me on Twitter.

Reverse a list in O(n)

2012-12-27T16:53:00-05:00

For about a year, I've always had a question I ask someone I'm interviewing for a programming position. I've asked this around 3 times and got some nice answers, but I had never implemented it. The question is really simple:

Given a linked list, how would you reverse it in O(n)?

Here's my solution:

Node = Struct.new(:id, :next) # Build the list first = Node.new(1, nil) nodes = 10.times.map{|x| Node.new(x + 2, nil)} nodes.each_with_index{|node, index| node.next = nodes[index + 1] } first.next = nodes.first def reverse_list(node) prox = node.next # Get the next node on the list. node.next = nil # Make the new tail point to nil (end of the new list). last = node # Save that new tail on a tmp variable (to point to it later) node = prox # Change the `current` node to the `next` one. Move ahead. while node # If I'm not at the end of the list. prox = node.next # Save the next node on the original list. node.next = last # Reverse the list (point back from `current`) last = node # Save the current node to point later. node = prox # Move ahead once. end last end def print_list(node) while node puts node.id node = node.next end end print_list(first) puts "\n" print_list(reverse_list(first))

How would you do it? There's a really simple solution using a stack, but if you find another one, I'd love to hear about it on Twitter @nhocki

Using Liquid `include` with DB templates

2012-09-05T17:37:00-04:00

I am working on a new application that will use some Liquid templates to allow our clients to edit some templates.

We will store these templates in the Database, load them, parse them and render them. So far, no magic.

But, I needed to have some "partials" for these. I wanted some common elements (like image galleries) to be available for the client to use in that specific template.

The way you normally accomplish this is by using an include tag, but I needed a little bit more control. Not all the clients will have the same partials and I also wanted the clients to be able to edit these partials.

So after reading the Liquid source code, I found a really easy way to do this.

You just need an object that responds to read_template_file and it can take 1 or 2 arguments (please, go for 2). This method must return the Liquid source of the partial (not a parsed template).

So, here's a really simple one that will work:

# Class to be used as a Liquid File System for the `include` tag class LiquidTemplateSystem attr_reader :user def initialize(user) @user = user end # Return a valid liquid template string def read_template_file(template_path, context) user.partials.find_by_name(template_path).liquid_source end end # In the view: # This `template` has an {% include %} tag. <%= template.render(obj, registers: LiquidTemplateSystem.new(user)) %>

Default FileSystem

You could also set the default FileSystem with Liquid::Template.file_system = MyNewFileSystem.new. This may be useful if you want the partials to be shared on the database but not different for every user.

Any other way?

Even though I find this to be really easy, another way may exist. If you know of a better way to do this, please let me know.

Seriously just do it

2012-07-26T01:07:00-04:00

I have finished my first "working season" at MIT. I got here on October 4. It's been an amazing and weird journey.

But I have to admit that at first I was a bad employee. I was really used to have somebody telling me what to do all the time. When I arrived at MIT that didn't happen. At all. I was given some "long term" tasks and that was it. I just had to do them.

Since no one was all over me checking my progress or asking me how I was doing on everything, I was waisting a lot (*most*) of my time. ESPNDeportes got a lot of visits from me. I started using Facebook on regular basis. Threadless didn't have enough T-shirts for me. Twitter was not "real time enough"…

But, with time, I started working and I can't say how grateful I am for this way of managing. I've learned to be much more productive on my own. To manage my time. To be much more responsible. But, I guess the best thing about this was that I learned that you just need to shut up, close all the meaningless things you have open and get shit done.

Sure, if you need help, ask. If what you're doing is wrong, try again. But you won't know something doesn't work if it's just an idea.

I became much more productive in everything. Not only MIT work stuff. Personal (disposable) things I wanted to try. Things I wanted to learn. Things I wanted to read (even my Instapaper queue went down. By a lot!).

So, just try it out. It may take more than a week to get used to it, but just try it. Close this post and every other distracting thing and do something.

Use the community for something more

2012-07-10T23:25:00-04:00

For a while, this idea has been wondering around in my head.

I would love to have some web-development community in Medellín, but I would like it to be "something more". I love the idea of learning from others, teaching and getting people involved. I know there are more people that are looking forward for this.

Sadly, we don't have a community like the one in Boston. It's amazing how many events there are here. Boston.rb meets up to 3 times a month! It's just crazy! You just need to go to Meetup to find an event of your interest! All filled with people eager to share and help!

But I would love to put all that lovin' to some "charity" work. Get real projects for people that need it. It doesn't have to be 100% free, but not charging nearly as much as you would normally do it!

For example, I know a Volunteer Fire Fighter Department that needs to build a simple learning system to distribute knowledge because ignorance is taking a lot of lives (from firefighters and people they're trying to save). Hell, they also need a new website!

I also know about a guy that needs a website for an organisation that helps small kids with cancer in Africa.

All these projects could use the expertise of "old" developers and enthusiasm from the ones that are just learning. I've found that working on a real project is way better than learning on your backyard. So why not take something that could use more man-power and pair program on these websites?

I'm not saying we should not have presentations! Presentations should still be the center of this. Sharing new stuff is the most important part for the community. But maybe having some hackathons where we work on these projects.

So, what do you think? Would something like this work? Would people be interested? Those organizations could, for example, sponsor the meetings giving some beer and food! Developing companies could also sponsor this and get some mention on the created websites. I know we would do it!

I can't do it right now. I'm 2.5K miles away! And I don't think it's an easy thing to organize, but I would like to know if people is actually interested before I throw myself sending emails and talking to people, so please contact me if you are, either by leaving a comment or pinging me on Twitter(@nhocki).

If you liked this idea, you might one to up vote it in Hackernews.

Fix Xcode Path

2012-05-09T19:38:00-04:00

I updated my Xcode version this week and to my surprise they moved it from /Developer to /Applications/Xcode, the problem is that now all the developer tools went missing. Here's the easy solution for this:

sudo /usr/bin/xcode-select -switch /Applications/Xcode.app/Contents/Developer/

That's it. Just let Xcode know where to look for the developer tools again.

Mixing Presenters and Helpers

2012-05-08T18:28:00-04:00

I just finished The Rails View and the best chapter for me was about presenters. I've been working with this for a while and I have to admit I love the approach presented on the view.

All the presenters I've done were instantiated in the controller. In the book they suggest that if the presenter doesn't depend on params or session variables it's the view responsibility to create it. This is extremely flexible specially when we're presenting various objects on the same view. Here's an example to present users in our application, let's use the same presenter for the index and show actions, but let's make it in a really simple way!

The Helper & View

Let's create the helper method first. It's really simple and will take a block that we can use to build a custom markup "on the fly".

# app/helpers/user_helpers.rb def basic_information_for(user = @user, options = {}) presenter = UserBasicPresenter.new(user, self, options) if block_given? yield(presenter) else presenter end end

Now, let's take a look at the views. In the show action, we're going to present only one user, and we can easily build the custom view for this. In the index action, we'll present all a "basic info" for the each user. We have that in a partial because this is used for all the collections of users in our application. Here's the view code:

# app/views/show.html.erb <% basic_information_for do |user| %> <%= user.name %> <%= user.status %> ... <% end %> # app/views/index.html <% @users.each do |user| %> <%= basic_information_for(user) %> <% end %>

The Presenter, A.K.A What's happening here?

What makes this really clean is that we can actually use <%= %> directly on our helper method. This works because we'll define the to_s method in our presenter! In this method we render our partial in the context of the helper, which is amazing. Here's the code for that:

# app/presenters/user_basic_presenter.rb class UserBasicPresenter def initialize(user, template, options = {}) @user = user @options = options @template = template end # more methods ... def to_s @template.render partial: 'user/basic', object: self end end

Is that simple! We have a unique interface (the helper) that will work great in both places! We could obviously make the presenter more complex, otherwise, why use it?

Even though I haven't used this approach a lot, I'm really happy with it. I felt the difference instantly and it made a lot of sense to me. Being able to mix this with a flexible layout is really powerful!

Feedback

I'm also really interested in how people are using presenters. I'd love to gain more feedbacks and points of view about this. Feel free to post a comment (here on in the gist) or ping me on Twitter, @nhocki.

Flexible and friendly layouts on rails

2012-03-29T23:16:00-04:00

I've been working on a "new" project (I'm new to the project, the project itself is 'old') and at the beginning I had to edit a bunch of stuff in the views.

A friend needed to change the header of the page on one action of the UsersController. He is new to rails and was struggling with this, so he asked me for help.

It's pretty common to have a header partial rendered on the view. If you're a lovely developer you would have render "header" and not something like render "layouts/header". The reason behind this is simple. Template Inheritance. This means that if you're working on the UsersController, when rails looks for the header partial, it will first look for users/header moving up the parents chain, ending in application/header. So, next time, put all your layout stuff in application/_partial so you can easily override it!

But this is a controller wide solution, since all the actions from that controller will render the same partial. So, how do you achieve this on a single action? My solution is a 3 line helper:

# This method will allow a section to be rendered form a view or # to render a partial instead. # # It's really useful to have 'flexible' sections in the layout. E.g # add this to your layout file: # # content_or_partial(:header) # => renders the content_for(:header) or the "header" partial # # content_or_partial(:header, 'users/hello') # => renders the content_for(:header) or the "users/hello" partial # # application.html.erb # <%= content_or_partial(:header) %> # # users/show.html.erb # <%= content_for :header do %> # My action specific content # <% end %> # def content_or_partial(content, partial = content) content_for?(content) ? content_for(content) : render(partial.to_s) end

Now all you have to do is fill your layouts with "content boxes" that can be changed on an controller or action-specific need, giving great flexibility.

How would you get around this?

Thoughts on Decorators & Presenters

2012-01-31T18:12:00-05:00

I have been developing a new app using a lot of decorators and presenters. I've been highly influenced by Avdi (both his blog and his Objects in Rails book) and Steve Klabnik, but now I have a problem. My decorators are starting to get big. I'll have to use concerns for that. Expect another post about it (maybe).

I want to warn you that I am not by any means an expert on the subject. So I'm more than open to any feedback you can give me.

All presenters are decorators, but not all decorators are presenters.

A decorator is a class that adds some sort of functionality to another class. A presenter is a class that adds some sort of presentation formatting functionality to another class.

Steve Klabnik

I think that what a decorator should do is give a standard API for the processed data stored somewhere (probably a database). I don't think a decorator should be in charge of the markup of the presented data. For me, it makes no sense that the decorator (that is processed data for me), knows how it is being shown to the user.

That said, I think the same decorator should be used to build the JSON response and the HTML response. The HTML markup and JSON structure are just \'markup\'. So, they can be treated as templates.

So, where do the logic in the template goes?

I had this discussion with a friend, and he asked me that, with this approach, how would he add elements to a view depending on a condition, for example, the famous \'***You are logged in as nhocki***\' if there's a user logged in.

I think this should be created inside helper methods that takes a block. Inside that block you can use the decorator methods to show the needed data. You can use those methods in your JSON builder too (try out the new Jbuilder). You could bundle up these methods and re-use them in your projects too.

Here's an example:

# app/models/user.rb # # User model has name & username # # Only the username is required (signup with Github for example) class User < ActiveRecord::Base validates :username, presence: true end # app/decorators/user_decorator.rb # # Let's use the Draper gem for the decorators class UserDecorator < ApplicationDecorator decorates :user def display_name user.name.presence || user.username end # Formatting the data is OK. I'm against adding the markup def salary h.number_to_currency(user.salary) end end # app/views/users/index.html.erb # current_user is a presenter <%= logged_in_area do %> You are logged in as <%= current_user.display_name %> <% end %> # viewing @some_users's profile (which is also a presenter) <%= admin_area do %> <%= @some_user.salary %> <% end %> # app/helpers/application_helpers.rb # # We all hate helpers, but this time, they're useful. def admin_area(&block) content_tag(:div, capture(&block), class: 'admin') if current_user.admin? end

Maybe I have the names wrong

As I mentioned before, I'm no expert on design patterns. And I'm pretty new to presenters and decorators. So, maybe my approach has a different name. I don't really care how this is called, I feel it's a more natural approach and it keeps the template and the processed data in different parts.

Please provide feedback

It is also possible that I have it all wrong, that's why I'm really interested in your opinion. I'm eager to learn about everything from anyone, so I would love to have some feedback about this. You can post a comment below or send me a message on twitter.

From Tumblr to Octopress

2012-01-23T00:00:00-05:00

I wrote an article yesterday about URL design and I spent about an hour fighting with Tumblr to format my code inside a pre (or code) tag. I knew I was done with them. Today, I'm giving Octopress a try. I honestly wanted to try Jekyll but I was really lazy to code my HTML. So, for now, I'm welcoming Octopress.

I really love the fact that I can write in Markdown, it's really powerful yet simple that makes me focus on writing and not formatting. I know I can use Markdown in Tumblr but I don't have Textmate there, so this is a big +1 for Jekyll/Octopress.

Moving Posts From Tumblr

There's a great script to move posts from Tumblr. They'll even be downloaded as markdown, but I still had to edit them to add the Gist snippets. Thankfully I don't have a lot of posts so it didn't took a lot.

Deploying

I've deployed my blog to Heroku now. It was supposed to be really simple. It was, after I realized that the public folder is ignored by the git. I spent an hour of my life there :-( !

Remove the public folder from your .gitignore

I'll probably post more about Octopress when I get to know it better. There are a bunch of plugins and stuff that may be useful, or maybe, it'll suck and I'll have to go and find something else!

Beautiful URLs in Rails. The easy way.

2012-01-22T00:00:00-05:00

I've always been in love with beautiful URL design. Specially for 'public' parts of the site. In Rails, it's pretty common to use the ID of the object in the URL. Well, that really sucks.

My first approach to solve this problem was started by Federico Builes a long time ago with a plugin (which I now maintain as a gem) called Make Permalink. It's really easy to use and will (up to some point) solve the problem. I know FriendlyID exists, but I hate having the permalink field in my database (I always forget it) so that's why I like Make Permalink's approach.

class Post < ActiveRecord::Base make_permalink :title # Override this method to have nice looking # URLs. You still use `Post.find` in the controller # and will get URL like this: # # /posts/1-look-mom-i-rock def to_param permalink end end

The good thing about having /users/1-nhocki as the URL is that is way better than /users/1 AND you don't have to change any of your app for that (you can still use find in your controllers).

I was really glad with this but not too long ago, someone asked me for something similar but for Mongoid. And hell! Mongoid ids are ugly as shit! So, they wanted to have just the username. I said, well, why don't you just change the to_param method to return the username and use User.find_by_username everywhere?

Even though this actually works, I would be very tired from writing find_by_username over and over again. So I told him to write a fetch method where he could get an user either by the id or the username. I think this is pretty convenient and nice, but I would hate to write that fetch method in every model. So, with some metaprogramming we can easily do this.

First, I've created a fetchable_on method to use it in your models. This will define a class method called fetch that will let you get the object by querying on the attribute OR on the id. Right now, it won't work if your attribute is a numeric value (like the number of bathrooms you have in your house), but most of the times, it's ok. Here's the code. Place it in your initializers folder.

# config/initializers/fetchable.rb # encoding: UTF-8 module Extensions module Fetchable # Adding `fetchable_on` to a model will define the `fetch` method # that will allow you to query that model with the `attribute` or `id`. # # Having the following: # # class User << ActiveRecord::Base # fetchable_on :username # end # # Will let you fetch users by username or id. # # User.fetch('nhocki').id # => 1 # User.fetch(1).username # => nhocki def fetchable_on(attribute) define_singleton_method("fetch") do |*args| looked_value = args.shift begin find(looked_value.to_i, *args) rescue ActiveRecord::RecordNotFound send("find_by_#{attribute}!", looked_value.to_s, *args) end end end end end ActiveRecord::Base.extend(Extensions::Fetchable)

That way, you just need to call fetchable_on :attribute in your model and it'll adjust the fetch method to query on that attribute or the id. Here's an example:

class User < ActiveRecord::Base fetchable_on :username # URL Style: /users/nhocki def to_param username end end # Gives you the `fetch` method. # It takes the username or the id. # Use fetch in your controller User.fetch('nhocki').id # => 1 User.fetch(1).username # => 'nhocki'

Note that I use find_by_attribute! to throw an exception if the object is not found. Using the method without a ! will return nil if there's no object in the DB.

How would you implement this? Any good solutions?

Simple Presenters & Decorators on Rails

2012-01-10T00:00:00-05:00

There's been a lot of talk about decorators and presenters (which are almost the same) in Rails and how they help you with a "real" OO code. There are great gems and there's even a Railscast (pro) on how to build a presenter from scratch. But, in the new project I'm working on, we (<- he) has this really simple setup which I feel kind of cleaner.

What we wanted to have was a present method in the controller and build the view from its methods. So, we created an ActiveSupport::Concern called Presentable and included it in our application controller. Here's our (once again, his) code for that module.

# app/concerns/controllers/presentable.rb # Presentable Concern # Include this module in the controllers where # you need the `present` method. module Presentable # This method returns a Presenter Object to work # with in the view. # # Attributes: # object: The object to present # # Supported Options # as: The presenter class. If this is not suplied # the method will try to guess the class # e.g: UserPresenter for a User object. def present(object, options = { }) if object.respond_to?(:first) return [] if object.empty? presenter = "#{object.first.class.to_s.pluralize}Presenter" else presenter = "#{object.class}Presenter" end presenter = options.fetch(:as) { eval(presenter) } presenter.new(object) end end

Now, we just need to create each presenter. But, since there's a lot of shared logic between presenters we (he) just created a parent Presenter class with all that. Notice that our presenter takes an object as parameter, so we just delegated all the method calls to our record (thank you method missing) if it responded to that method. You could also send the view context to access Rails helpers inside your presenter. Here's our Presenter base class.

# Base presenter class # All presenters should inherit from this one. class Presenter # Include all the modules you need or # send the view context to access rails helpers. include Rails.application.routes.url_helpers # ... include Presentable attr_reader :record def initialize(record) @record = record @klass = record.class end def method_missing(name, *arguments, &block) super unless record.respond_to?(name) record.send(name, *arguments, &block) end def translate_field_name(name) @klass.human_attribute_name(name) end # UrlHelper was being an ass, easier to go commando. def link_to(name, url, options={}) href = { :href => url } content_tag(:a, name, options.merge(href)) end end

Then, we just need to inherit from the Presenter class and add our custom, clean methods. Using delegate and method missing will help us keep the class clean. I won't show you a presenter here, they'll look a lot like Ryan's one in his Railscasts so check that out first.

I will show you our controller though. Using the decent_exposure gem we managed to get super tiny controllers. Here's 1 of them:

class RequestsController < ApplicationController expose(:filter) { params.fetch(:filter) { :all} } expose(:filtered_requests) { Request.filter(filter) } expose(:requests) { present(filtered_requests) } before_filter :find_request, :only => [:show, :destroy] def index end def show extend(RequestRedirector) # Another concern redirect_to_request end def destroy @request.destroy redirect_to requests_path(:filter => "new_requests"), :notice => "Se ha eliminado la solicitud" end private def find_request @request = Request.find(params[:id]) end end

Combining presenters and form builders will make your views even cleaner, giving a lot of pleasure to the future you! So please, just pretty-please, start using presenters and builders and leave helpers as simple as possible. Or, leave them for good.

You'll be amazed how different you'll feel when reading some old code or some co-worker's code.

But free web services are not like free software

2011-12-06T00:00:00-05:00

But free web services are not like free software. If your free software project suddenly gets popular, you gain resources: testers, developers and people willing to pitch in. If your free website takes off, you lose resources. Your time is spent firefighting and your money all goes to the nice people at Linode.

Don't be a Free User - Pinboard Blog

Colorize your rake output

2011-11-30T00:00:00-05:00

I've been writing a lot of rake tasks lately. I like to have a lot of output (feedback) from my tasks, at least while I'm writing them. I have colors in my tests, but my output is really ugly, so I took some code from Chatte, - a simple chat application a friend (and awesome developer, btw) wrote for college- and made some modifications for it. Here's my module right now…

module Colors def colorize(text, color_code) "\033[#{color_code}m#{text}\033[0m" end { :black => 30, :red => 31, :green => 32, :yellow => 33, :blue => 34, :magenta => 35, :cyan => 36, :white => 37 }.each do |key, color_code| define_method key do |text| colorize(text, color_code) end end end # green "Hello" # => "\e[32mHello\e[0m"

Now, you just need to include Colors in your tasks and enjoy your life.

I know there's a colorize gem out there, but I don't like their syntax. I prefer Andres' approach much more, so maybe I'll turn this into a really simple gem. Do you think it's worth it?

Algorithms are on the cutting edge of entrepreneurship.

2011-11-09T00:00:00-05:00

Algorithms are on the cutting edge of entrepreneurship. If you're talking about just re-implementing stuff that people did ten years ago, performance isn't that important at some level. But, if you're talking about doing stuff that nobody has done before, one of the reasons often that they haven't done it is because it's too time-consuming.

By Charles Leiserson

Compile assets and push them - no pain involved!

2011-09-28T00:00:00-04:00

This week, I've been working on a side project with Andrés Mejía (@andmej) and for some really really sad reason Heroku is not working as expected (a simple git push should do the trick, but we're having a really weird issue).

At first, I thought it had something to do with the assets. Even-though Heroku has asset compiling tasks on their Cedar stack, it didn't really worked for us at first. So I just decided to create a small rake task to compile the assets and push them to Github.

The good thing about this rake task is that it will only push the assets. This means, that you'll need a clean git tree to run this task or it will die. This ensures that the automatic push won't commit any non-related change to the compilation of the assets.

Now, you only have to run rake assets:compile and sleep well at night. Here's the code. Hope it helps.

namespace :assets do desc "Compile all the assets named in config.assets.precompile and push them" task :compile do AssetsCompiler.new.compile end class AssetsCompiler def compile ensure_clean_git removed_previous_assets compile_assets commit_compiled_assets push end def ensure_clean_git raise "Can't deploy without a clean git status." if git_dirty? end def removed_previous_assets run "bundle exec rake RAILS_ENV=production assets:clean" end def compile_assets run "bundle exec rake RAILS_ENV=production assets:precompile" end def commit_compiled_assets run "git add -u && git add . && git commit -am 'Compiled assets'" end def push run "git push" end private def run(command) puts "+ Running: #{command}" puts "-- #{system command}" end def git_dirty? `[[ $(git diff --shortstat 2> /dev/null | tail -n1) != "" ]]` dirty = $?.success? end end end

Simple scrollbar synchronization with jQuery

2011-09-21T00:00:00-04:00

I'm working on a really cool project on my University. It's a web application to compare book editions and help professors build what is know as the Collatzio. The Collatzio is simply the "original" version of the book. That means, to have the book the author wanted to write and not the one that the editors wanted to publish.

I needed to have two versions of the book aligned side by side. So, the HTML/CSS was pretty simple. Just have two simple divs with overflow-y: auto; to make the scrollbar appears on each div.

The coffeescript code I used is really simple, here it is:

jQuery ($) -> ($ ".scroll-text").scroll (event) -> trigger = ($ @) element = trigger.data("scrolls") $(element).scrollTop(trigger.scrollTop()) event.preventDefault() return false

Hope it helps someone. I struggled a little bit with the "common" way of scrolling things. People usually use the animate method to scroll the body, but I couldn't use it because it kept triggering the scroll event on the elements, so it was an infinite loop.

$("html, body").animate({ scrollTop: 0 }, "slow");

TimeHub, the first winner!

2011-08-26T00:00:00-04:00

I just wanted to "personally" say thanks to everyone who voted for my entry on the Rally on Rails (Latinamerican Rails Rumble).

Timehub won! And this couldn't be possible without the people who voted.

I would like to thank the Rally on Rails team & judges. They made an excellent job and even though there are somethings to improve (like the voting rules), I think that the contest was great!

We had already posted a "thank you note", in the Timehub blog, but I really wanted to write one myself.

I was pretty spammer during these last 3 days! That will stop now!

Thanks again! Hope to see you in Timehub v2!