Skip to content
Feature-rich testing framework for Crystal based on RSpec.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
spec Update spec with new arguments Aug 17, 2019
.gitlab-ci.yml Cache bin/ directory for ameba Apr 7, 2019


Spectator is a fully-featured spec-based test framework for Crystal. It provides more functionality from RSpec than the built-in Crystal Spec utility. Additionally, Spectator provides extra features to make testing easier and more fluent.


Spectator is designed to:

  • Reduce complexity of test code.
  • Remove boilerplate from tests.
  • Lower the difficulty of writing non-trivial tests.
  • Provide an elegant syntax that is easy to read and understand.
  • Provide common utilities that the end-user would otherwise need to write.


Add this to your application's shard.yml:

    gitlab: arctic-fox/spectator


If it doesn't exist already, create a spec/ file. In it, place the following:

require "spectator"
require "../src/*"

This will include Spectator and the source code for your shard. Now you can start writing your specs. The syntax is the same as what you would expect from modern RSpec. The "expect" syntax is recommended and the default, however the "should" syntax is also available. Your specs must be wrapped in a Spectator.describe block. All other blocks inside the top-level block may use describe and context without the Spectator. prefix.

Here's a minimal spec to demonstrate:

require "./spec_helper"

Spectator.describe String do
  subject { "foo" }

  describe "#==" do
    context "with the same value" do
      let(value) { subject.dup }

      it "is true" do eq(value)

    context "with a different value" do
      let(value) { "bar" }

      it "is false" do
        is_expected.to_not eq(value)

If you find yourself trying to shoehorn in functionality or unsure how to write a test, please create an issue for it. The goal is to make it as easy as possible to write specs and keep your code clean. We may come up with a solution or even introduce a feature to support your needs.

NOTE: Due to the way this shard uses macros, you may find that some code you would expect to work, or works in other spec libraries, creates syntax errors. If you run into this, please create an issue so that we may try to resolve it.


Spectator has all of the basic functionality for BDD. For full documentation on what it can do, please visit the wiki.


The DSL supports arbitrarily nested contexts. Contexts can have values defined for multiple tests (let and subject). Additionally, hooks can be used to ensure any initialization or cleanup is done (before, after, and around). Pre- and post-conditions can be used to ensure code contracts are kept.

# Initialize the database before running the tests in this context.
before_all { Database.init }

# Teardown the database and cleanup after tests in the is context finish.
after_all { Database.cleanup }

# Before each test, add some rows to the database.
let(row_count) { 5 }
before_each do
  row_count.times { Database.insert_row }

# Remove the rows after the test to get a clean slate.
after_each { Database.clear }

describe "#row_count" do
  it "returns the number of rows" do
    expect(Database.row_count).to eq(row_count)

Spectator has different types of contexts to reduce boilerplate. One is the sample context. This context type repeats all tests (and contexts within) for a set of values. For instance, some feature should behave the same for different input. However, some inputs might cause problems, but should behave the same. An example is various strings (empty strings, quoted strings, strings with non-ASCII, etc), and numbers (positive, negative, zero, NaN, infinity).

# List of integers to test against.
def various_integers
  [-7, -1, 0, 1, 42]

# Repeat nested tests for every value in `#various_integers`.
sample various_integers do |int|
  # Example that checks if a fictitious method `#format` converts to strings.
  it "formats correctly" do
    expect(format(int)).to eq(int.to_s)

Another context type is given. This context drastically reduces the amount of code needed in some scenarios. It can be used where one (or more inputs) changes the output of multiple methods. The given context gives a concise syntax for this use case.

subject(user) { }

# Each expression in the `given` block is its own test.
given age = 10 do
  expect(user.can_drive?).to be_false
  expect(user.can_vote?).to be_false

given age = 16 do
  expect(user.can_drive?).to be_true
  expect(user.can_vote?).to be_false

given age = 18 do
  expect(user.can_drive?).to be_true
  expect(user.can_vote?).to be_true


Spectator supports two formats for assertions (expectations). The preferred format is the "expect syntax". This takes the form:

expect(THIS).to eq(THAT)

The other format, "should syntax" is used by Crystal's default Spec.

THIS.should eq(THAT)

The first format doesn't monkey-patch the Object type. And as a bonus, it captures the expression or variable passed to expect(). For instance, compare these two tests:

foo = "Hello world"
foo.size.should eq(12) # Wrong on purpose!

Produces this error output:

Failure: 11 does not equal 12

  expected: 11
    actual: 12

Which is reasonable, but where did 11 come from? Alternatively, with the "expect syntax":

foo = "Hello world"
expect(foo.size).to eq(12) # Wrong on purpose!

Produces this error output:

Failure: foo.size does not equal 12

  expected: 12
    actual: 11

This makes it clearer what was being tested and failed.


Spectator has a variety of matchers for assertions. These are named in such a way to help tests read as plain English. Matchers can be used on any value or block.

There are typical matchers for testing equality: eq and ne. And matchers for comparison: <, <=, >, >=, be_within. There are matchers for checking contents of collections: contain, have, start_with, end_with, be_empty, have_key, and more. See the wiki for a full list of matchers.


Spectator supports multiple options for running tests. "Fail fast" aborts on the first test failure. "Fail blank" fails if there are no tests. Tests can be filtered by their location and name. Additionally, tests can be randomized. Spectator can be configured with command-line arguments, a config block in a file, and .spectator config file.

Spectator.configure do |config|
  config.fail_blank # Fail on no tests.
  config.randomize  # Randomize test order.
  config.profile    # Display slowest tests.


Spectator matches Crystal's default Spec output with some minor changes. JUnit and TAP are also supported output formats. There is also a highly detailed JSON output.


This shard is still under development and is not recommended for production use (same as Crystal). However, feel free to play around with it and use it for non-critical projects.

Feature Progress

In no particular order, features that have been implemented and are planned. Items not marked as completed may have partial implementations.

  • DSL
    • describe and context blocks
    • Contextual values with let, let!, subject, described_class
    • Test multiple and generated values - sample, random_sample
    • Concise syntax - given
    • Before and after hooks - before_each, before_all, after_each, after_all, around_each
    • Pre- and post-conditions - pre_condition, post_condition
    • Other hooks - on_success, on_failure, on_error
    • One-liner syntax
    • Should syntax - should, should_not
    • Helper methods and modules
    • Aliasing - custom example group types with preset attributes
    • Pending tests - pending
    • Shared examples - behaves_like, include_examples
  • Matchers
    • Equality matchers - eq, ne, be ==, be !=
    • Comparison matchers - be <, be <=, be >, be >=, be_within[.of], be_close
    • Type matchers - be_a, respond_to
    • Collection matchers
      • contain
      • have
      • contain_exactly
      • contain_exactly.in_any_order
      • match_array
      • match_array.in_any_order
      • start_with
      • end_with
      • be_empty
      • have_key
      • have_value
      • all
      • all_satisfy
    • Truthy matchers - be, be_true, be_truthy, be_false, be_falsey, be_nil
    • Error matchers - raise_error
    • Yield matchers - yield_control[.times], yield_with_args[.times], yield_with_no_args[.times], yield_successive_args
    • Output matchers - output[.to_stdout|.to_stderr]
    • Predicate matchers - be_x, have_x
    • Misc. matchers
      • match
      • satisfy
      • change[.by|.from[.to]|.to|.by_at_least|.by_at_most]
      • have_attributes
    • Compound - and, or
  • Runner
    • Fail fast
    • Test filtering - by name, context, and tags
    • Fail on no tests
    • Randomize test order
    • Dry run - for validation and checking formatted output
    • Config block in
    • Config file - .spectator
  • Reporter and formatting
    • RSpec/Crystal Spec default
    • JSON
    • JUnit
    • TAP

How it Works (in a nutshell)

This shard makes extensive use of the Crystal macro system to build classes and modules. Each describe and context block creates a new module nested in its parent. The it block creates an example class. An instance of the example class is created to run the test. Each example class includes a context module, which contains all test values and hooks.


  1. Fork it (
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Merge Request

Please make sure to run crystal tool format before submitting. The CI build checks for properly formatted code. Ameba is run to check for code style.

Tests must be written for any new functionality. Macros that create types are not as easy to test, so they are exempt for the current time. However, please test all code locally with an example spec file.

Documentation is automatically generated and published to GitLab pages. It can be found here:

This project is developed on GitLab, and mirrored to GitHub. Issues and PRs/MRs are accepted on both.


You can’t perform that action at this time.