Fork me on GitHub


A tiny debugging library for Ruby


If you think about it, Letters is a library dedicated to side effects. It wants to stay out of your logic, to keep clear of application state. The purpose of Letters is to give you a window into an otherwise closed pipeline or code path, and to make that window as easy as possible to set up and tear down.

Letters shouldn’t stay in your code base, so it takes the liberty of patching Object. This happens when you enter:

require "letters"

For Rails apps, there may be conflicts with the t method in ActionPack. To handle that conflict, do not require all of Letters. Instead, do two things:

# In your Gemfile
          gem "letters", :require => false
          # In application.rb after Bundler.require
          require "letters/patch/rails"

If you don’t want to patch everything, you can patch classes and objects a la carte:

require "letters/patch"
          Letters.patch! Hash
          obj =
          Letters.patch! obj

A (assert)

The a method stands apart from most other Letters methods. Why? Because it does not always return its receiver. a is used to craft assertions, and if an assertion fails, it raises an error. Two special cases of the a method are e and n, which assert that their receivers are not empty or nil (respectively).

The a method jumps into the context of its receiver, so all assertions can be made without explicitly referring to the object:

[1, 2, 3].a { count == 3 }
          # => [1, 2, 3]

Notice that you didn’t have to pass a block argument in.

Assertions will fail if the result of the block is not truthy (as defined by Ruby). That is, false or nil blocks will raise a Letters::AssertionError.

[1, 2, 3].a { count > 3 }
          # Raises Letters::AssertionError
          [1, 2, 3].a { nil }
          # Raises Letters::AssertionError

The a method can take a message that will be printed every time the assertion is made. It can also take an Exception class to be raised if the assertion fails.


:message => nil
          :error_class => Letters::AssertionError

B (beep)

The b method causes your terminal to, well, beep. This one is mainly for fun but is also useful for coarse-grained time analysis. For example, you might leverage b to detect n + 1 SQL queries or other data manipulation that takes more than a few milliseconds.

          # => 5000000150000000

If your terminal supports the audible bell, this query will beep every time the pipeline reaches a b and will pass through the results to the next method.

For more fine-grained time analysis, see the timestamp method, t.

C (callstack)

The c method prints the current callstack.

def inner
            # Print the callstack
            rand.c * 10
          def outer(multiplier)
            inner * multiplier
          outer 4

In IRB, this prints:

          <main>:1:in `inner_method'
          <main>:5:in `outer_method'
          (irb):3:in `irb_binding' 
            <rubydir>/irb/workspace.rb:80: in `eval'

Again, this does not interrupt execution of your code – it just lets you know where you are.


:message => nil

D (debugger)

We all know and love the debugger, but normal debugger calls are imperative. It can be a pain to break apart code to fit in a debugger. Instead, consider invoking d at the end of any expression:

Network.fetch_data.d.all? do |str|
            str.length > 25 

The d method will be even more powerful when combined with other constructs like transmitters and receivers (not yet implemented).

D1/D2 pairs (object diff)

The d1/d2 method pair brings diff to your Ruby environment. Instead of printing a line-by-line diff, though, this diff prints out a hash. How does this method pair work? All you need to do is mark the first object for comparison with d1 and then call d2 on the second object. By default, d1 and d2 work on arrays, hashes, and any object that defines a sane minus method (-) for comparing itself with like objects.

[1, 2, 3] {|x| x ** 2 }.d2

Calling the expression will print the following in the Awesome Print format:

            removed: [2, 3],
            added: [4, 9]

For hashes, each value in the diff is also a hash (and there is an updated key):

{ a: "foo", b: "bar", c: "baz" } do |k, v|
            k =~ /[ab]/
          end.merge(:a => "new-foo", d: "bat").d2
          # => { a: "new-foo", b: "bar" }
          # Prints:
          # {
          #   added: { d: "bat" }
          #   removed: { c: "baz" }
          #   updated: { a: "new-foo" }
          # }

Options for d2

:message => nil
          :format => "ap"
          :stream => $stdout

E (empty check)

The e method is meant to quickly check if an expression is empty. One of three methods that does not always return its receiver, e will raise a Letters::EmptyError if its receiver is empty. (The other methods are k and n.) Of course, if it is not empty, the receiver is passed through.

          # => raises Letters::EmptyError
          [1, 2, 3].e
          # => [1, 2, 3]


:message => nil

F (write to file)

Sometimes, you want to be able to manipulate the results of a method call in your favorite text editor. Maybe you want to use sophisticated Unix tools to pick apart an unnecessarily gigantic object. You could always stop to break open your code, create a file block, remember the file flag options, and accidentally change downstream code in the process. Or you could tag your object with f:

JSON.parse(body).f.values_at(:name, :title)

By default, this will write to a file called “log” in your current directory. It will dump the object out in YAML format. To change either of those settings, add parameters to f:

object.f(:format => "json")
          object.f(:name => "file.txt")


:format => "yaml"
          :name => "log"

See “Formats” below for all available formats.

J (jump into object)

The j method gives you a block in the context of the object it’s called on. You can call any of the object’s methods (in addition to puts, etc., for finer-grained debugging) without naming a receiver. (You could be explicit and use self. But what does this look like, Python?) Note that if you mutate the object, it will be mutated on the other side of the method.

[1, 2, 3].j { puts length unless empty? }.reduce(:+)
          # => 6

This expression will print 3 and return 6.

K (kill)

One of three methods that does not always return its receiver, k raises a Letters::KillError when called too many times. The k command can be great for debugging recursion or to check out infinite loops. (The other methods are e and n.) It can also be used to stop running your code at any point to check the state of the world outside your code (for example in a database).

def recurse(hash, count=0)
            print count
            recurse(hash.k(:on => 3), count + 1)
          # Prints 0123
          # => raises Letters::KillError
          [1, 2, 3].k(:on => 1)
          # => [1, 2, 3]


:on => 0

L (logger)

The l method assumes you have an instance of a Ruby logger returned by the method logger. This will be the case in any standard Rails or Sinatra app. You can also set one up using the standard Ruby logger or Log4r.


:format => "yaml"
          :level => "info"

M (mark as tainted, untainted)

While not used every day, tainting and untainting objects gives us more control over what we allow through our code. In Ruby, tainted objects (mostly) represent user input and derived values. They can be tainted and untainted at will at the lower safety levels.

m can taint or untaint its receiver object. Without an argument, m will taint its receiver. With a falsy argument, m will untaint its receiver.

[1, 2, 3].m.o { tainted? }
          # => [1, 2, 3]
          # Prints "true"
          [1, 2, 3].m(true).m(false).o { tainted? }
          # => [1, 2, 3]
          # Prints "false"

N (nil check)

The n method is meant to quickly check if an expression is nil. One of two methods that does not always return its receiver, n will raise a Letters::NilError if its receiver is nil. (The other such method is e, used to check for empty objects.) Of course, if it is not nil, the receiver is passed through.

          # => raises Letters::NilError
          [1, 2, 3].n
          # => [1, 2, 3]

O (print to STDOUT)

Note: The o method was formerly the p method. I changed it in case people want to use Letters alongside other debugging methods, like Kernel#p

O prints to $STDOUT. You know how to use this one.

[1, 2, 3].o
          # => [1, 2, 3]

By default, this will print the object to STDOUT in Awesome Print format. To change the format, add a parameter to o:

[1, 2, 3].o(:format => "yaml")

If a block is passed in, it will be executed “inside” of the object (like the j method), and the final result of the block will be printed out instead.

The j example looked something like this:

[1, 2, 3].j { puts length unless empty? }.reduce(:+)

With o, we could rewrite the expression as …

[1, 2, 3].o { length unless empty? }.reduce(:+)

… for the same effect.


:format => "ap"
          :stream => $stdout

R (RI)

You’ve probably forgotten about RI, haven’t you? It’s that tool that comes with Ruby and is meant to make offline documentation easy.

Because there are plenty of resources on the Internet, namely RubyDoc, people tend to disable RDoc generation. But context-switching from the terminal/keyboard to the browser/mouse can be disruptive. So Letters gives you the power to explore Ruby’s documentation from the comfort of your own terminal.

To check out the documentation for an object’s class, simply use the r method:

[1, 2, 3].r
          # => [1, 2, 3]

When you call this method, you will get the following in STDOUT:

          = Array < Object
          = Includes:
          Enumerable (from ruby site)
          (from ruby site)
          Arrays are ordered, integer-indexed collections of
          any object. Array indexing starts at 0, as in C or
          Java. A negative index is assumed to be relative
          to the end of the array---that is, an index of -1
          indicates the last element of the array, -2 is the
          next to last element in the array, and so on.
          = Class methods:
            [], new, try_convert
          ... etc., etc. ...

Not interested in learning what an array is? Pass in a method name:

[1, 2, 3].r(:grep)
          # => [1, 2, 3]

Now you get more focused information:

          = Array#grep
          (from ruby site)
          === Implementation from Enumerable
            enum.grep(pattern)                   -> array
            enum.grep(pattern) {| obj | block }  -> array
          Returns an array of every element in enum for
          which Pattern === element. If the optional block
          is supplied, each matching element is passed to
          it, and the block's result is stored in the output
            (1..100).grep 38..44   #=> [38, 39, 40, 
                                        41, 42, 43, 44]
            c = IO.constants
            c.grep(/SEEK/)         #=> [:SEEK_SET,
            res = c.grep(/SEEK/) {|v| IO.const_get(v) }
            res                    #=> [0, 1, 2]

If you’re using RVM and need to generate your RDoc again, type the following in and go grab a coffee:

          rvm docs generate all

S (bump safety level)

Ruby’s safety level is sort of like the US national security level. With an escalated safety level, it’s harder to change things that should be easy, and nothing gets done. Really, the only difference between Ruby’s security level and the United States' is that Ruby’s does not default to Threat Level Orange.

Though not often used, the safety level can be an englightening tool for debugging, both to gauge the attack vectors your code is vulnerable to, and to see how much user input you’re relying on.

For more information on the Ruby safety level, see the online Pickaxe documentation.

The s method will bump the safety level up by one. If supplied a specific number (0 - 4), it will try to change the safety level to that number. If the safety level cannot be changed, this method will raise an error.

The s method is most interesting when combined with tainted objects.

T (timestamp)

The t method will print out the current timestamp. This can be useful for identifying bottlenecks in code more precisely than with b but without the complexity of a profiler.

          # => 5000000150000000

This call prints something like:

          09/24/2012 13:35:15.282
          09/24/2012 13:35:33.016
          09/24/2012 13:35:43.172


The :time_format option allows you to print timestamps in any time format you’ve registered with ActiveSupport.

:time_format => "millis"


The following formats are supported. They can be specified by passing format: "format" to appropriate methods.

  • Ruby Pretty Print (format: "pp")
  • Ruby Awesome Print (format: "ap")
  • YAML (format: "yaml")
  • JSON (format: "json")
  • XML (format: "xml")

*Requiring "letters" on its own will add the alphabet methods to Object.

To patch just the core classes, require "letters/patch/core".

If you don’t want to patch them with such small method names, you can explicitly require "letters/patch" instead. Letters.patch! will be available to patch any class or instance you like.

If you do patch an instance, the letter methods will only be available on that instance and not on any derivative instances. For example, this will not work:

require "letters/patch"
          arr = [1, 2, 3]
          Letters.patch! arr
          # Does not work
 {|x| x ** 2 }.o

The second call to o will fail because the derivative array (result of the map call) has not been patched. Of course, mutating the original array is one way to solve this problem, albeit error-prone:

# Works, but patching Array is probably better
! {|x| x ** 2 }.o