branch: elpa/buttercup
commit 512b82b22ff3fe98baff23d93ef8733850558864
Author: Jorgen Schaefer <cont...@jorgenschaefer.de>
Commit: Jorgen Schaefer <cont...@jorgenschaefer.de>
Rework documentation to make the readme more concise.
Fixes #12 by adding an "why not ERT?" section.
---
Makefile | 2 +-
README.md | 684 ++++----------------------------------
docs/images/buttercup.jpg | Bin 0 -> 21461 bytes
README.md => docs/introduction.md | 97 +-----
docs/testrunners.md | 15 +
docs/usage.md | 129 +++++++
6 files changed, 221 insertions(+), 706 deletions(-)
diff --git a/Makefile b/Makefile
index 2192f61..8c82578 100644
--- a/Makefile
+++ b/Makefile
@@ -8,7 +8,7 @@ DISTFILES = buttercup.el buttercup-compat.el buttercup-pkg.el
README.md
all: test
test:
- $(EMACS) -batch -L . -l buttercup.el -f buttercup-run-markdown README.md
+ $(EMACS) -batch -L . -l buttercup.el -f buttercup-run-markdown
docs/introduction.md
./bin/buttercup -L .
compile:
diff --git a/README.md b/README.md
index 5eaa3b8..7c5a662 100644
--- a/README.md
+++ b/README.md
@@ -3,31 +3,22 @@
[](https://travis-ci.org/jorgenschaefer/emacs-buttercup)
[](http://stable.melpa.org/#/buttercup)
+
+
Buttercup is a behavior-driven development framework for testing Emacs
-Lisp code. It is heavily inspired by
-[Jasmine](https://jasmine.github.io/). So heavily inspired, in fact,
-that half this page is more or less a verbatim copy of the
-[Jasmine introduction](https://jasmine.github.io/edge/introduction.html).
+Lisp code. It allows to group related tests so they can share common
+set-up and tear-down code, and allows the programmer to “spy” on
+functions to ensure they are called with the right arguments during
+testing.
-All code in this file can be run by Buttercup’s built-in markdown test
-runner. Just use `make test` in the project directory to see the
-output.
+The framework is heavily inspired by
+[Jasmine](https://jasmine.github.io/edge/introduction.html).
-## Getting Started
+## Example
-You can install buttercup from
-[MELPA Stable](http://stable.melpa.org/). Add the following to your
-`init.el` or `.emacs` file:
+*Full article: [Introduction](docs/introduction.md)*
-```
-(require 'package)
-(add-to-list 'package-archives
- '("melpa-stable" . "http://stable.melpa.org/packages/";) t)
-```
-
-This should allow you to `M-x package-install RET buttercup RET`.
-Afterwards, in an otherwise empty directory, create a file called
-`some-test.el` with the following contents:
+A simple test looks like this.
```Lisp
(describe "A suite"
@@ -35,55 +26,23 @@ Afterwards, in an otherwise empty directory, create a file
called
(expect t :to-be t)))
```
-You can now use buttercup to run this test:
-
-```
-$ emacs -batch -f package-initialize -l buttercup -f buttercup-run-discover
-A suite
- contains a spec with an expectation
-```
-
-Congratulations, you just ran your first tests!
-
-## Full Project Example
-
-Tests should be run in an environment separate from your interactive
-Emacs. To get such an environment and install packages into it, you
-can use [Cask](https://github.com/cask/cask). For installation
-instructions see there.
-
-In the directory from above, create a new file called `Cask` with the
-following contents:
-
-```
-(source gnu)
-(source melpa-stable)
-
-(development
- (depends-on "buttercup"))
-```
+## Installation and Usage
-Now you can tell cask to download and install the depended-on packages
-in the environment of the current directory:
+*Full article: [Usage](docs/usage.md)*
-```
-$ cask install
-```
-
-You only need to do this once. Subsequently, you can run buttercup
-tests simply by invoking buttercup using cask:
+You can install buttercup from
+[MELPA Stable](http://stable.melpa.org/). Add the following to your
+`init.el` or `.emacs` file:
```
-$ cask exec buttercup
-A suite
- contains a spec with an expectation
+(require 'package)
+(add-to-list 'package-archives
+ '("melpa-stable" . "http://stable.melpa.org/packages/";) t)
```
-## Suites: `describe` Your Tests
+This should allow you to `M-x package-install RET buttercup RET`.
-A test suite begins with a call to the Buttercup macro `describe` with
-the first parameter describing the suite and the rest being the body
-of code that implements the suite.
+Now create a file called `test-feature.el` with these contents:
```Lisp
(describe "A suite"
@@ -91,560 +50,55 @@ of code that implements the suite.
(expect t :to-be t)))
```
-## Specs
-
-Specs are defined by calling the Buttercup macro `it`, which, like
-`describe` takes a string and code. The string is the title of the
-spec and the code is the spec, or test. A spec contains one or more
-expectations that test the state of the code. An expectation in
-Buttercup is an assertion that is either true or false. A spec with
-all true expectations is a passing spec. A spec with one or more false
-expectations is a failing spec.
-
-### It’s Just Functions
-
-The code arguments to `describe` and `it` is just turned into
-functions internally, so they can contain any executable code
-necessary to implement the rules. Emacs Lisp scoping rules apply, so
-make sure to define your spec file to be lexically scoped.
-
-```Lisp
-(describe "A suite is just a function"
- (let ((a nil))
- (it "and so is a spec"
- (setq a t)
- (expect a :to-be t))))
-```
-
-## Expectations
-
-Expectations are expressed with the `expect` function. Its first
-argument is the actual value. The second argument is a test, followed
-by expected values for the test to compare the actual value against.
-
-If there is no test, the argument is simply tested for being non-nil.
-This can be used by people who dislike the matcher syntax.
-
-### Matchers
-
-Each matcher implements a boolean comparison between the actual value
-and the expected value. It is responsible for reporting to Buttercup
-if the expectation is true or false. Buttercup will then pass or fail
-the spec.
-
-Any matcher can evaluate to a negative assertion by prepending it with
-the `:not` matcher.
-
-```Lisp
-(describe "The :to-be matcher compares with `eq'"
- (it "and has a positive case"
- (expect t :to-be t))
- (it "and can have a negative case"
- (expect nil :not :to-be t)))
-```
-
-### Included Matchers
-
-Buttercup has a rich set of matchers included. Each is used here — all
-expectations and specs pass. There is also the ability to write custom
-matchers (see the `buttercup-define-matcher` macro for further
-information) for when a project’s domain calls for specific assertions
-that are not included below.
-
-```Lisp
-(describe "Included matchers:"
- (it "The :to-be matcher compares with `eq'"
- (let* ((a 12)
- (b a))
- (expect a :to-be b)
- (expect a :not :to-be nil)))
-
- (describe "The :to-equal matcher"
- (it "works for simple literals and variables"
- (let ((a 12))
- (expect a :to-equal 12)))
-
- (it "should work for compound objects"
- (let ((foo '((a . 12) (b . 34)))
- (bar '((a . 12) (b . 34))))
- (expect foo :to-equal bar))))
-
- (it "The :to-match matcher is for regular expressions"
- (let ((message "foo bar baz"))
- (expect message :to-match "bar")
- (expect message :to-match (rx "bar"))
- (expect message :not :to-match "quux")))
-
- (it "The :to-be-truthy matcher is for boolean casting testing"
- (let (a
- (foo "foo"))
- (expect foo :to-be-truthy)
- (expect a :not :to-be-truthy)))
-
- (it "The :to-contain matcher is for finding an item in a list"
- (let ((a '("foo" "bar" "baz")))
- (expect a :to-contain "bar")
- (expect a :not :to-contain "quux")))
-
- (it "The :to-be-less-than matcher is for mathematical comparisons"
- (let ((pi 3.1415926)
- (e 2.78))
- (expect e :to-be-less-than pi)
- (expect pi :not :to-be-less-than e)))
-
- (it "The :to-be-greater-than matcher is for mathematical comparisons"
- (let ((pi 3.1415926)
- (e 2.78))
- (expect pi :to-be-greater-than e)
- (expect e :not :to-be-greater-than pi)))
-
- (it "The :to-be-close-to matcher is for precision math comparison"
- (let ((pi 3.1415926)
- (e 2.78))
- (expect pi :not :to-be-close-to e 2)
- (expect pi :to-be-close-to e 0)))
-
- (describe "The :to-throw matcher"
- (it "is for testing if a function throws an exception"
- (let ((foo (lambda () (+ 1 2)))
- (bar (lambda () (+ a 1))))
- (expect foo :not :to-throw)
- (expect bar :to-throw)))
- (it "accepts a symbol to check for the signal thrown"
- (let ((foo (lambda () (/ 1 0)))
- (bar (lambda () (+ a 1))))
- (expect foo :not :to-throw 'void-variable)
- (expect bar :to-throw 'void-variable)))
- (it "optionally matches arguments to signals"
- (let ((foo (lambda () (+ a 1)))
- (bar (lambda () (+ a 1))))
- (expect foo :not :to-throw 'void-variable '(b))
- (expect bar :to-throw 'void-variable '(a))))))
-```
-
-## Grouping Related Specs with `describe`
-
-The `describe` macro is for grouping related specs. The string
-parameter is for naming the collection of specs, and will be
-concatenated with specs to make a spec’s full name. This aids in
-finding specs in a large suite. If you name them well, your specs read
-as full sentences in traditional
-[BDD](http://en.wikipedia.org/wiki/Behavior-driven_development) style.
-
-```Lisp
-(describe "A spec"
- (it "is just a function, so it can contain any code"
- (let ((foo 0))
- (setq foo (1+ foo))
-
- (expect foo :to-equal 1)))
-
- (it "can have more than one expectation"
- (let ((foo 0))
- (setq foo (1+ foo))
-
- (expect foo :to-equal 1)
- (expect t :to-equal t))))
-```
-
-### Setup and Teardown
-
-To help a test suite DRY up any duplicated setup and teardown code,
-Buttercup provides the `before-each`, `after-each`, `before-all` and
-`after-all` special forms.
-
-As the name implies, code blocks defined with `before-each` are called
-once before each spec in the `describe` is run, and the `after-each`
-code blocks are called once after each spec.
-
-Here is the same set of specs written a little differently. The
-variable under test is defined at the top-level scope — the `describe`
-block — and initialization code is moved into a `before-each` block.
-The `after-each` block resets the variable before continuing.
-
-```Lisp
-(describe "A spec using `before-each' and `after-each'"
- (let ((foo 0))
- (before-each
- (setq foo (1+ foo)))
-
- (after-each
- (setq foo 0))
-
- (it "is just a function, so it can contain any code"
- (expect foo :to-equal 1))
-
- (it "can have more than one expectation"
- (expect foo :to-equal 1)
- (expect t :to-equal t))))
-```
-
-The `before-all` form is called only once before all the specs in
-`describe` are run, and the `after-all` form is called after all specs
-finish. These functions can be used to speed up test suites with
-expensive setup and teardown.
-
-However, be careful using `before-all` and `after-all`! Since they are
-not reset between specs, it is easy to accidentally leak state between
-your specs so that they erroneously pass or fail.
-
-```Lisp
-(describe "A spec using `before-all' and `after-all'"
- (let (foo)
- (before-all
- (setq foo 1))
-
- (after-all
- (setq foo 0))
-
- (it "sets the initial value of foo before specs run"
- (expect foo :to-equal 1)
- (setq foo (1+ foo)))
-
- (it "does not reset foo between specs"
- (expect foo :to-equal 2))))
-```
-
-### Nesting `describe` Blocks
-
-Calls to `describe` can be nested, with specs defined at any level.
-This allows a suite to be composed as a tree of functions. Before a
-spec is executed, Buttercup walks down the tree executing each
-`before-each` function in order. After the spec is executed, Buttercup
-walks through the `after-each` functions similarly.
-
-```Lisp
-(describe "A spec"
- (let (foo)
- (before-each
- (setq foo 0)
- (setq foo (1+ foo)))
-
- (after-each
- (setq foo 0))
-
- (it "is just a function, so it can contain any code"
- (expect foo :to-equal 1))
-
- (it "can have more than one expectation"
- (expect foo :to-equal 1)
- (expect t :to-equal t))
-
- (describe "nested inside a second describe"
- (let (bar)
- (before-each
- (setq bar 1))
-
- (it "can reference both scopes as needed"
- (expect foo :to-equal bar))))))
-```
-
-## Disabling Suites
-
-Suites and specs can be disabled with the `xdescribe` and `xit`
-macros, respectively. These suites and any specs inside them are
-skipped when run and thus their results will not appear in the
-results.
-
-```Lisp
-(xdescribe "A spec"
- (let (foo)
- (before-each
- (setq foo 0)
- (setq foo (1+ foo)))
-
- (it "is just a function, so it can contain any code"
- (expect foo :to-equal 1))))
-```
-
-## Pending Specs
-
-Pending specs do not run.
-
-Any spec declared with `xit` is marked as pending.
-
-Any spec declared without a function body will also be marked as
-pending in results.
-
-```Lisp
-(describe "Pending specs"
- (xit "can be declared using `xit'"
- (expect t :to-be nil))
-
- (it "can be declared with `it' but without a body"))
-```
-
-## Spies
-
-Buttercup has test double functions called spies. While other
-frameworks call these mocks and similar, we call them spies, because
-their main job is to spy in on function calls. Also, Jasmine calls
-them spies, and so do we. A spy can stub any function and tracks calls
-to it and all arguments. A spy only exists in the `describe` or `it`
-block it is defined in, and will be removed after each spec. There are
-special matchers for interacting with spies. The
-`:to-have-been-called` matcher will return true if the spy was called
-at all. The `:to-have-been-called-with` matcher will return true if
-the argument list matches any of the recorded calls to the spy.
-
-```Lisp
-(describe "A spy"
- (let (foo bar)
- (before-each
- (setf (symbol-function 'foo)
- (lambda (value)
- (setq bar value)))
-
- (spy-on 'foo)
-
- (foo 123)
- (foo 456 "another param"))
-
- (it "tracks that the spy was called"
- (expect 'foo :to-have-been-called))
-
- (it "tracks all arguments of its calls"
- (expect 'foo :to-have-been-called-with 123)
- (expect 'foo :to-have-been-called-with 456 "another param"))
-
- (it "stops all execution on a function"
- (expect bar :to-be nil))))
-```
-
-### Spies: `:and-call-through`
-
-The keyword argument `:and-call-through` to `spy-on` will make the spy
-call the original function instead of returning `nil`.
-
-```Lisp
-(describe "A spy, when configured to call through"
- (let (bar set-bar get-bar fetched-bar)
- (before-each
- (fset 'set-bar (lambda (val)
- (setq bar val)))
- (fset 'get-bar (lambda ()
- bar))
-
- (spy-on 'get-bar :and-call-through)
-
- (set-bar 123)
- (setq fetched-bar (get-bar)))
-
- (it "tracks that the spy was called"
- (expect 'get-bar :to-have-been-called))
-
- (it "should not affect other functions"
- (expect bar :to-equal 123))
-
- (it "when called returns the requested value"
- (expect fetched-bar :to-equal 123))))
-```
-
-### Spies: `:and-return-value`
-
-The keyword argument `:and-return-value` specifies the value the
-spied-on function should return.
-
-```Lisp
-(describe "A spy, when configured to fake a return value"
- (let (bar set-bar get-bar fetched-bar)
- (before-each
- (fset 'set-bar (lambda (val)
- (setq bar val)))
- (fset 'get-bar (lambda ()
- bar))
-
- (spy-on 'get-bar :and-return-value 745)
-
- (set-bar 123)
- (setq fetched-bar (get-bar)))
-
- (it "tracks that the spy was called"
- (expect 'get-bar :to-have-been-called))
-
- (it "should not affect other functions"
- (expect bar :to-equal 123))
-
- (it "when called returns the requested value"
- (expect fetched-bar :to-equal 745))))
-```
-
-### Spies: `:and-call-fake`
-
-The keyword argument `:and-call-fake` delegates calls to a supplied
-function.
-
-```Lisp
-(describe "A spy, when configured with an alternate implementation"
- (let (bar set-bar get-bar fetched-bar)
- (before-each
- (fset 'set-bar (lambda (val)
- (setq bar val)))
- (fset 'get-bar (lambda ()
- bar))
-
- (spy-on 'get-bar :and-call-fake (lambda () 1001))
-
- (set-bar 123)
- (setq fetched-bar (get-bar)))
-
- (it "tracks that the spy was called"
- (expect 'get-bar :to-have-been-called))
-
- (it "should not affect other functions"
- (expect bar :to-equal 123))
-
- (it "when called returns the requested value"
- (expect fetched-bar :to-equal 1001))))
-```
-
-### Spies: `:and-throw-error`
-
-With the keyword argument `:and-throw-error`, all calls to the spy
-will `signal` the specified value as an error.
-
-```Lisp
-(describe "A spy, when configured to throw an error"
- (let (bar set-bar get-bar fetched-bar)
- (before-each
- (fset 'set-bar (lambda (val)
- (setq bar val)))
- (fset 'get-bar (lambda ()
- bar))
-
- (spy-on 'get-bar :and-throw-error 'error))
-
- (it "throws the error"
- (expect (lambda () (get-bar))
- :to-throw 'error))))
-```
-
-### Other tracking properties
-
-Every call to a spy is tracked and exposed using the `spy-calls`
-accessor.
-
-`spy-calls-any` returns `nil` if the spy has not been called at all,
-and then `t` once at least one call happens. `spy-calls-count` returns
-the number of times the spy was called. `spy-calls-args-for` returns
-the arguments passed to a given call (by index). `spy-calls-all-args`
-returns the arguments to all calls. `spy-calls-all` returns the
-current buffer and arguments passed to all calls.
-`spy-calls-most-recent` returns the current buffer and arguments for
-the most recent call. `spy-calls-first` returns the current buffer and
-arguments for the first call.
-
-Finally, `spy-calls-reset` clears all tracking for a spy.
-
-```Lisp
-(describe "A spy"
- (let (set-foo foo)
- (before-each
- (fset 'set-foo (lambda (val &rest ignored)
- (setq foo val)))
- (spy-on 'set-foo))
-
- (it "tracks if it was called at all"
- (expect (spy-calls-any 'set-foo)
- :to-equal
- nil)
-
- (set-foo 5)
-
- (expect (spy-calls-any 'set-foo)
- :to-equal
- t))
-
- (it "tracks the number of times it was called"
- (expect (spy-calls-count 'set-foo)
- :to-equal
- 0)
-
- (set-foo 2)
- (set-foo 3)
-
- (expect (spy-calls-count 'set-foo)
- :to-equal
- 2))
-
- (it "tracks the arguments of each call"
- (set-foo 123)
- (set-foo 456 "baz")
-
- (expect (spy-calls-args-for 'set-foo 0)
- :to-equal
- '(123))
-
- (expect (spy-calls-args-for 'set-foo 1)
- :to-equal
- '(456 "baz")))
-
- (it "tracks the arguments of all calls"
- (set-foo 123)
- (set-foo 456 "baz")
-
- (expect (spy-calls-all-args 'set-foo)
- :to-equal
- '((123)
- (456 "baz"))))
-
- (it "can provide the context and arguments to all calls"
- (set-foo 123)
-
- (expect (spy-calls-all 'set-foo)
- :to-equal
- `(,(make-spy-context :current-buffer (current-buffer)
- :args '(123)
- :return-value nil))))
-
- (it "has a shortcut to the most recent call"
- (set-foo 123)
- (set-foo 456 "baz")
-
- (expect (spy-calls-most-recent 'set-foo)
- :to-equal
- (make-spy-context :current-buffer (current-buffer)
- :args '(456 "baz")
- :return-value nil)))
-
- (it "has a shortcut to the first call"
- (set-foo 123)
- (set-foo 456 "baz")
-
- (expect (spy-calls-first 'set-foo)
- :to-equal
- (make-spy-context :current-buffer (current-buffer)
- :args '(123)
- :return-value nil)))
-
- (it "can be reset"
- (set-foo 123)
- (set-foo 456 "baz")
-
- (expect (spy-calls-any 'set-foo)
- :to-be
- t)
-
- (spy-calls-reset 'set-foo)
+You can now use buttercup to run this test:
- (expect (spy-calls-any 'set-foo)
- :to-be
- nil))))
```
+$ emacs -batch -f package-initialize -f buttercup-run-discover
+Running 1 specs.
+A suite
+ contains a spec with an expectation
-## Test Runners
-
-Evaluating `describe` forms just stores the suites. You need to use a
-test runner to actually evaluate them. Buttercup comes with two test
-runners by default:
-
-- `buttercup-run-at-point` — Evaluate the topmost `describe` form at
- point and run the suite it creates directly. Useful for interactive
- development. But be careful, this uses your current environment,
- which might not be clean (due to said interactive development).
-- `buttercup-run-discover` — Find files in directories specified on
- the command line, load them, and then run all suites defined
- therein. Useful for being run in batch mode.
-- `buttercup-run-markdown` — Run code in markdown files. Used to
- run this file’s code.
+Ran 1 specs, 0 failed, in 0.0 seconds.
+```
+
+Congratulations, you ran your first test!
+
+## Feature List
+
+- Shared set-up and tear-down sections to reduce code repetition and
+ share a common environment among tests.
+- Easy to read and extensible `expect` macro to describe expected
+ behavior.
+- Powerful mocking framework, called “spies,” to both cause them to
+ return expected values or throw errors as needed by the test, as
+ well as to ensure functions are called with expected arguments
+ during tests.
+- Built to be run within a Continuous Integration environment,
+ including test runners to discover and execute tests in a directory
+ tree.
+
+### Why not ERT?
+
+Emacs comes with a testing framework,
+[ERT](https://www.gnu.org/software/emacs/manual/html_mono/ert.html).
+Buttercup was written to address some shortcomings of that framework.
+
+- ERT
+ [deliberately leaves it up to the programmer to define set-up and tear-down
code](https://www.gnu.org/software/emacs/manual/html_mono/ert.html#Fixtures-and-Test-Suites),
+ which requires a lot of boiler-plate code for every set-up function.
+ Buttercup makes this easy and seamless.
+- ERT has no good way of being run in a continuous integration
+ environment. There are
+ [external projects to make this less of a
pain](https://github.com/rejeep/ert-runner.el)
+ instead. Once all is said and done, you installed six external
+ packages your project does not need just to run your own tests. And
+ that does not include a mocking library.
+- ERT has no way of grouping related tests, requiring every test name
+ to share the same prefix, making names long and hard to read.
+
+Nonetheless, ERT is a great project. It introduced testing to Emacs,
+and Buttercup learned a lot from its code to record a stack trace for
+error display. Even though Buttercup tries to be a better testing
+framework than ERT, we do wish ERT and the ERT maintainers all the
+best and hope both frameworks can continue to benefit from each other.
diff --git a/docs/images/buttercup.jpg b/docs/images/buttercup.jpg
new file mode 100644
index 0000000..f6966b4
Binary files /dev/null and b/docs/images/buttercup.jpg differ
diff --git a/README.md b/docs/introduction.md
similarity index 85%
copy from README.md
copy to docs/introduction.md
index 5eaa3b8..6e2938d 100644
--- a/README.md
+++ b/docs/introduction.md
@@ -1,84 +1,18 @@
-# Buttercup — Behavior-Driven Emacs Lisp Testing
-
-[](https://travis-ci.org/jorgenschaefer/emacs-buttercup)
-[](http://stable.melpa.org/#/buttercup)
+# Introduction
Buttercup is a behavior-driven development framework for testing Emacs
-Lisp code. It is heavily inspired by
-[Jasmine](https://jasmine.github.io/). So heavily inspired, in fact,
-that half this page is more or less a verbatim copy of the
+Lisp code. It does not depend on any other Emacs Lisp libraries. It
+has a clean, obvious syntax so that you can easily write tests.
+
+It is heavily inspired by [Jasmine](https://jasmine.github.io/). So
+heavily inspired, in fact, that most of this page is more or less a
+verbatim copy of the
[Jasmine introduction](https://jasmine.github.io/edge/introduction.html).
All code in this file can be run by Buttercup’s built-in markdown test
runner. Just use `make test` in the project directory to see the
output.
-## Getting Started
-
-You can install buttercup from
-[MELPA Stable](http://stable.melpa.org/). Add the following to your
-`init.el` or `.emacs` file:
-
-```
-(require 'package)
-(add-to-list 'package-archives
- '("melpa-stable" . "http://stable.melpa.org/packages/";) t)
-```
-
-This should allow you to `M-x package-install RET buttercup RET`.
-Afterwards, in an otherwise empty directory, create a file called
-`some-test.el` with the following contents:
-
-```Lisp
-(describe "A suite"
- (it "contains a spec with an expectation"
- (expect t :to-be t)))
-```
-
-You can now use buttercup to run this test:
-
-```
-$ emacs -batch -f package-initialize -l buttercup -f buttercup-run-discover
-A suite
- contains a spec with an expectation
-```
-
-Congratulations, you just ran your first tests!
-
-## Full Project Example
-
-Tests should be run in an environment separate from your interactive
-Emacs. To get such an environment and install packages into it, you
-can use [Cask](https://github.com/cask/cask). For installation
-instructions see there.
-
-In the directory from above, create a new file called `Cask` with the
-following contents:
-
-```
-(source gnu)
-(source melpa-stable)
-
-(development
- (depends-on "buttercup"))
-```
-
-Now you can tell cask to download and install the depended-on packages
-in the environment of the current directory:
-
-```
-$ cask install
-```
-
-You only need to do this once. Subsequently, you can run buttercup
-tests simply by invoking buttercup using cask:
-
-```
-$ cask exec buttercup
-A suite
- contains a spec with an expectation
-```
-
## Suites: `describe` Your Tests
A test suite begins with a call to the Buttercup macro `describe` with
@@ -631,20 +565,3 @@ Finally, `spy-calls-reset` clears all tracking for a spy.
:to-be
nil))))
```
-
-
-## Test Runners
-
-Evaluating `describe` forms just stores the suites. You need to use a
-test runner to actually evaluate them. Buttercup comes with two test
-runners by default:
-
-- `buttercup-run-at-point` — Evaluate the topmost `describe` form at
- point and run the suite it creates directly. Useful for interactive
- development. But be careful, this uses your current environment,
- which might not be clean (due to said interactive development).
-- `buttercup-run-discover` — Find files in directories specified on
- the command line, load them, and then run all suites defined
- therein. Useful for being run in batch mode.
-- `buttercup-run-markdown` — Run code in markdown files. Used to
- run this file’s code.
diff --git a/docs/testrunners.md b/docs/testrunners.md
new file mode 100644
index 0000000..34aae59
--- /dev/null
+++ b/docs/testrunners.md
@@ -0,0 +1,15 @@
+# Test Runners
+
+Evaluating `describe` forms just stores the suites. You need to use a
+test runner to actually evaluate them. Buttercup comes with three test
+runners by default:
+
+- `buttercup-run-at-point` — Evaluate the topmost `describe` form at
+ point and run the suite it creates directly. Useful for interactive
+ development. But be careful, this uses your current environment,
+ which might not be clean (due to said interactive development).
+- `buttercup-run-discover` — Find files in directories specified on
+ the command line, load them, and then run all suites defined
+ therein. Useful for being run in batch mode.
+- `buttercup-run-markdown` — Run code in markdown files. Used to
+ run this file’s code.
diff --git a/docs/usage.md b/docs/usage.md
new file mode 100644
index 0000000..e6be162
--- /dev/null
+++ b/docs/usage.md
@@ -0,0 +1,129 @@
+# Usage
+
+Buttercup is primarily meant to be used non-interactively, to
+automatically test a project independent of a user’s setup, before a
+commit and on a continuous integration platform. Because of this, the
+workflow for buttercup does not use interactive commands but instead
+the command line.
+
+## Cask
+
+[Cask](https://github.com/cask/cask) is a project to create an
+environment separate from your usual interactive Emacs environment.
+This allows you to install the packages your project depends on, and
+only those, and test your project in a well-defined environment.
+
+Buttercup works best in such environments, so the following best
+practices rely on Cask to be installed.
+
+## Project Directory Layout
+
+A basic project layout requires a project file, called `feature.el`
+here, a `Cask` file to define dependencies, and a `tests/` directory
+for tests. It should look roughly like this:
+
+```
+feature/feature.el
+ Cask
+ tests/feature.el
+```
+
+**feature.el**
+
+```Lisp
+(defun featurize (bug feature)
+ (format "It's not a %s, it's a %s" bug feature))
+
+(provide 'feature)
+```
+
+**Cask**
+
+```
+(source gnu)
+(source melpa-stable)
+
+(development
+ (depends-on "buttercup"))
+```
+
+**tests/test-feature.el**
+
+```Lisp
+(require 'feature)
+
+(describe "The feature"
+ (it "can use bug and feature"
+ (expect (featurize "bug" "feature")
+ :to-equal
+ "It's not a bug, it's a feature")))
+```
+
+## Running Tests
+
+You can now use Cask to run your tests.
+
+First, you have to install the dependencies. You only have to do this
+once, or when the dependencies change:
+
+```
+$ cask
+Extracting buttercup-1.1/
+Extracting buttercup-1.1/buttercup-compat.el
+Extracting buttercup-1.1/buttercup.el
+Extracting buttercup-1.1/bin/
+Extracting buttercup-1.1/bin/buttercup
+Extracting buttercup-1.1/buttercup-pkg.el
+Generating autoloads for buttercup-compat.el...
+Generating autoloads for buttercup-compat.el...done
+Generating autoloads for buttercup-pkg.el...
+Generating autoloads for buttercup-pkg.el...done
+Generating autoloads for buttercup.el...
+Generating autoloads for buttercup.el...done
+```
+
+Now, you can run your tests:
+
+```
+$ cask exec buttercup -L .
+Running 1 specs.
+
+The feature
+ can use bug and feature
+
+Ran 1 specs, 0 failed, in 0.0 seconds.
+```
+
+That’s it.
+
+You can run this command whichever way you like. Common choices
+include a makefile or shell scripts.
+
+## Travis
+
+If your project is hosted on github, you can use
+[Travis CI](https://travis-ci.org/) as your continuous integration
+environment. Buttercup can easily be used in such a setup. Simply add
+the following `.travis.yml` file:
+
+```
+language: emacs-lisp
+env:
+ - EVM_EMACS=emacs-24.4-bin
+ - EVM_EMACS=emacs-24.5-bin
+before_install:
+ - sudo mkdir /usr/local/evm
+ - sudo chown $(id -u):$(id -g) /usr/local/evm
+ - curl -fsSkL https://raw.github.com/rejeep/evm/master/go | bash
+ - export PATH="$HOME/.evm/bin:$PATH"
+ - evm install $EVM_EMACS --use
+ - curl -fsSkL https://raw.github.com/cask/cask/master/go | python
+ - export PATH="$HOME/.cask/bin:$PATH"
+ - cask
+script:
+ - cask exec buttercup -L .
+```
+
+Most of the complexity here is from installing
+[EVM](https://github.com/rejeep/evm) and Cask to be able to test your
+project using different Emacs versions.
