The most significant improvements over CoffeeScript.

I count tokens in many examples below, to show how CaffeineScript improves over CoffeeScript and JavaScript ES6. Learn more about Counting Tokens.

Core Improvements

(I'm factoring this page out into multiple, focused pages.)

• Blocks Instead of Brackets
• Extract
• Strings
• Array and Object Literals
• CommonJs
• Include
• Binary Operator Line Starts

Block method invocation

Invoke a method by following it with an indented list of arguments on separate lines:

# CaffeineScript
method
  arg1
  arg2
  arg3
  arg4

# CoffeeScript
method arg1,
  arg2
  arg3
  arg4

# Javascript
method(arg1, arg2, arg3, arg4);

Other Improvements / Refinements

Auto 'do'

Do you ever have bugs because you forgot a "do"? I do all the time. I think defaulting to 'do' will cause less bugs.

(This needs performance testing and semantic refinement.)

# CoffeeScript
for a in b
  do (a) ->
    -> a

# CaffeineScript
# for-block is an implicit closure
# If a function is created in the for-block, wrap all variables
# in a closure so each iteration gets its own copy.
# Only capture variables first-assigned in the for-block and not used outside.
for a in b
  -> a

Smart '@' binding

Basically, make '->' work more consistently with other languages. Most the time we shouldn't have to think about '@' binding for instance variables. Only in rare occasions should we have to override the default.

• In class definitions and not inside other explicit functions, -> works like CoffeeScript
• Inside an explicit function definitions, -> binds '@' to its enclosing scope. (like => in CoffeeScript)
• Overrides
  • ~> is always unbounded (-> in CoffeeScript)
  • => is always bounded (just like in CoffeeScript)

# CoffeeScript
class Foo extends Art.Foundation.BaseObject
  @setter
    baz: (a) -> ...

  bar: (list) ->
    list.each (a) =>
      @baz = a

# CaffeineScript
class Foo extends Art.Foundation.BaseObject
  @setter
    baz: (a) -> ...

  bar: (list) ->
    list.each (a) ->
      @baz a

map, each and for loops

CoffeeScript's 'for' loop is actually a 'map' operation. Sometimes you don't want that. Unwanted creation and population of an array is both a performance problem and a semantic problem.

# CoffeeScript
# create and return new array
(list) ->
  for element in list
    element.foo

# return list; doesn't create a new array
(list) ->
  for element in list
    @add element
  list

# CaffeineScript
# create and return new array
(list) ->
  array element from list
    element.foo

# return list; doesn't create a new array
(list) ->
  each element in list
    @add element

Extended unquoted labels:

# CoffeeScript
foo: 1
'foo.bar': 2
'foo-bar': 3

# CaffeineScript
foo: 1
foo.bar: 2
foo-bar: 3