• Ruby performance: #attr_accessor vs. method definition

    February 19, 2018

    I’ve been digging a lot into Ruby performance lately and this much digging has taken me into some very interesting corners of Rubyland. My latest surprise has been the difference between defining attribute methods via the attr_accessor provided language construct (as well as attr_reader and attr_writer) vs. defining them yourself (as in def attribute and def attribute=). Here’s what I ran into…

    I created two simple classes with identical interfaces. The only difference between the two is how we are defining access to set and get the @value attribute in each:

    class TestClassAttrAccessor
      attr_accessor :value
      def initialize value
        @value = value
    class TestClassDefMethod
      def initialize value
        @value = value

    Continue Reading →

  • Ruby performance: Hash's #has_key? vs. #[] (square brackets)

    February 15, 2018

    During my time last year developing performance improvements for Rambling Trie, I stumbled into something quite interesting that happens with Ruby’s Hash class.

    A commonly used method from the Hash interface is #has_key? which tells you if the Hash in question contains a particular key. Rambling Trie’s underlying data structure is an n-ary tree backed by a Hash where each key is the letter corresponding to a child and each value is the Node that corresponds to that letter. As you might imagine, #has_key? is a common operation called throughout the gem implementation.

    To my surprise, while running some Ruby benchmarks I noticed that accessing the key with #[] and verifying if it was nil instead of calling #has_key? lowered the time it took…

    Continue Reading →

  • Ruby performance: Using benchmarks and memory_profiler to chase performance degradations

    February 11, 2018

    While working on some of the performance improvements for version 1.0.0 of Rambling Trie late in 2016 and early in 2017, I upgraded from Ruby 2.3.3 to Ruby 2.4.0 imagining that the newer version would help me on my mission to make all operations execute a bit faster. To my surprise, I ran into a significant performance degradation after upgrading - specifically, there was a ~25% increase in the time it took to execute a script using a trie with a big-ish word dictionary. This intrigued me enough that I decided to take a closer look.

    Benchmarking with Ruby

    After some investigation with the help of Ruby’s own Benchmark, I realized that while most operations were a bit slower, the main problem was a lot more visible during intensive operations…

    Continue Reading →

  • Rambling Trie 1.0.0 released!

    January 23, 2017

    It’s been a while since I last wrote anything about Rambling Trie (or anything at all really), but I wanted to take the time to talk about all the changes and performance improvements that I’ve made to the gem lately and that culminated in the release of version 1.0.0.

    The changes worth highlighting are:

    NEW: Ability to dump and load tries from disk

    It takes a relatively long time to load all the words from a large dictionary into a new trie in memory. This becomes especially annoying when you have to do this full process every time you restart your application, even though you know that the trie is going to remain constant over time.

    To sort this out, you can now use the Rambling::Trie.load and Rambling::Trie.dump methods! Like this:

    Continue Reading →

  • Specifying failure message for RSpec expectations

    April 12, 2016

    When writing the Unit Testing ActiveRecord eager-loading blog post a couple months ago, I noticed that the test failures I was getting while writing were not very helpful. Initially, I had an assertion that looked like this:

    expect(restaurant.association(:reviews)).to be_loaded

    Since I was writing the test before the implementation — *cough* TDD isn’t dead *cough* — I got the expected failure when running it. However, the failure message was not that friendly after all:

    $ rspec spec/controllers/restaurant_controller_spec.rb
      1) RestaurantsController#index eager loads
         Failure/Error: expect(restaurant.association(:reviews)).to be_loaded
           expected `#<ActiveRecord::Associations::HasManyAssociation (...) >.loaded?` to return…

    Continue Reading →

  • Page Objects: Where do you put your assertions?

    April 12, 2016

    I started writing about Page Objects earlier this month, and as soon as you start talking about Page Objects, you need to have an opinion about where to put your assertions. Do you ask the page if it is in the state it expects to be or do you ask the page for values and assert that they are what you expect them to be in the test?

    If you look at FluentLenium and Simplelenium on GitHub, you’ll notice that their corresponding READMEs describe the usage of the Page Object pattern. Now, if you take a closer look, you’ll also notice that they have an opinion: they’ve included the assertions in the definition of such Page Objects.

    Now, my first experience with Page Objects was actually under these conditions. The Page Objects were the main driver…

    Continue Reading →

  • Using Page Objects in your Acceptance Tests

    April 10, 2016

    If you’ve written acceptance tests for web applications in the past (also called feature tests), you might be familiar with tools like Capybara, Simplelenium and FluentLenium. These are great abstractions over the browser (thanks, Selenium!) that provide very nice APIs for testing web applications.

    If you’ve done this for a while, you might also have heard of Page Objects. The idea behind them is that your tests should be about the behavior of your application and not about the underlying HTML, since the HTML is an implementation detail and probably not the interesting part of your tests.

    Our base acceptance test

    Let’s assume that we are working on an application where you can browse and review restaurants and we have an acceptance test…

    Continue Reading →

  • Unit Testing ActiveRecord eager-loading

    January 27, 2016

    If you’ve worked with relational databases and any ORMs like Java’s Hibernate, .NET’s NHibernate or Rails’ ActiveRecord in the past, you might be familiar with SELECT N+1 issues. It is a common performance problem in database-dependent applications and, because of this, these ORMs provide a built-in solution to this problem.

    In ActiveRecord, includes, preload and eager_load come to the rescue. Therefore, it is not unusual to find these keywords scattered in different places where your application accesses the database. Hopefully this isn’t a lot of places though - you are using Query Objects, right?

    An example application

    Let’s imagine for a second that we have an application where you can browse restaurants, which in turn have many reviews…

    Continue Reading →

  • Query Objects in the Rails world - A Different Approach

    January 26, 2016

    If you have worked with Ruby on Rails before then you might be familiar with ActiveRecord scopes. Using them, you can achieve what many would consider very readable code. Let’s say that we have an application where we display an inbox where users receive messages.

    class Message < ActiveRecord::Base

    Now, let’s imagine that after reading a Message, it is marked as read, and let’s represent that with a read column in the database. Additionally, our users can either archive the Message or move it to the trash. We’ll represent this concept with a location column in the messages table.

    Querying the database the Rails way

    Let’s say that our users want to have a way to view unread messages in their inbox. Using ActiveRecord, you could achieve…

    Continue Reading →

  • Installing MacVim with Lua enabled through Homebrew

    January 11, 2016

    If you’re like me, you like to have control over the plugins that you have installed in your editor setup. However, staying up to date with the different plugins can be a little bit of a hassle from time to time, so lately, I’ve been using YADR, a set of community maintained dotfiles.

    A great autocompletion plugin that YADR comes with is neocomplete.vim, which “provides a keyword completion system by maintaining a cache of keywords in the current buffer”. Unfortunately, for it to work, it needs MacVim with Lua enabled.

    Now, I already had MacVim installed through Caskroom:

    which mvim
    # /usr/local/bin/mvim
    ls -l $(which mvim)
    # /usr/local/bin/mvim -> /opt/homebrew-cask/Caskroom/macvim/7.4-84/mvim

    But this version was installed without Lua…

    Continue Reading →