Haml and Sass are templating engines for the two most common types of documents on the web: HTML and CSS, respectively. They are designed to make it both easier and more pleasant to code HTML and CSS documents, by eliminating redundancy, reflecting the underlying structure that the document represents, and providing elegant, easily understandable, and powerful syntax.
There are several ways to use Haml and Sass. They can be used as a plugin for Rails or Merb, or embedded on their own in other applications. The first step of all of these is to install the Haml gem:
gem install haml
To install Haml and Sass as a Rails plugin, just run haml --rails path/to/rails/app and both Haml and Sass will be installed. Views with the .haml (or .html.haml for edge) extension will automatically use Haml. Sass is a little more complicated; .sass files should be placed in public/stylesheets/sass, where they’ll be automatically compiled to corresponding CSS files in public/stylesheets when needed (the Sass template directory is customizable… see the Sass module docs for details).
For Merb, .html.haml views will work without any further modification. To enable Sass, you also need to add a dependency. To do so, just add
dependency "merb-haml"
to config/dependencies.rb (or config/init.rb in a flat/very flat Merb application). Then it’ll work just like it does in Rails.
To use Haml and Sass programatically, check out the RDocs for the Haml and Sass modules.
The most basic element of Haml is a shorthand for creating HTML tags:
%tagname{:attr1 => 'value1', :attr2 => 'value2'} Contents
No end-tag is needed; Haml handles that automatically. Adding class and id attributes is even easier. Haml uses the same syntax as the CSS that styles the document:
%tagname#id.class
In fact, when you’re using the <div> tag, it becomes even easier. Because <div> is such a common element, a tag without a name defaults to a div. So
#foo Hello!
becomes
<div id='foo'>Hello!</div>
Haml uses indentation to bring the individual elements to represent the HTML structure. A tag’s children are indented beneath than the parent tag. Again, a closing tag is automatically added. For example:
%ul %li Salt %li Pepper
becomes:
<ul> <li>Salt</li> <li>Pepper</li> </ul>
You can also put plain text as a child of an element:
%p Hello, World!
It’s also possible to embed Ruby code into Haml documents. An equals sign, =, will output the result of the code. A hyphen, -, will run the code but not output the result. You can even use control statements like if and while:
%p
Date/Time:
- now = DateTime.now
%strong= now
- if now > DateTime.parse("December 31, 2006")
= "Happy new " + "year!"
Haml provides far more tools than those presented here. Check out the reference documentation in the Haml module.
At its most basic, Sass is just another way of writing CSS. Although it’s very much like normal CSS, the basic syntax offers a few helpful features: tabulation indicates the attributes in a rule, rather than non-DRY brackets; and newlines indicate the end of an attribute, rather than a semicolon. For example:
#main :background-color #f00 :width 98%
becomes:
#main {
background-color: #f00;
width: 98% }
However, Sass provides much more than a way to make CSS look nice. In CSS, it’s important to have accurate selectors, so your styles don’t just apply to everything. However, in order to do this, you need to use nested element selectors. These get very ugly very quickly. I’m sure everyone’s had to write something like “#main .sidebar .top p h1 a”, followed by “#main .sidebar .top p h1 a:visited” and “#main .sidebar .top p h1 a:hover”. Well, Sass gets rid of that. Like Haml, it uses indentation to indicate the structure of the document. So, what was:
#main {
width: 90%;
}
#main p {
border-style: solid;
border-width: 1px;
border-color: #00f;
}
#main p a {
text-decoration: none;
font-weight: bold;
}
#main p a:hover {
text-decoration: underline;
}
becomes:
#main
:width 90%
p
:border-style solid
:border-width 1px
:border-color #00f
a
:text-decoration none
:font-weight bold
a:hover
:text-decoration underline
Pretty nice, no? Well, it gets better. One of the main complaints against CSS is that it doesn’t allow variables. What if have a color or a width you re-use all the time? In CSS, you just have to re-type it each time, which is a nightmare when you decide to change it later. Not so for Sass! You can use the “!” character to set variables. Then, if you put “=” after your attribute name, you can set it to a variable. For example:
!note_bg= #55aaff
#main
:width 70%
.note
:background-color= !note_bg
p
:width 5em
:background-color= !note_bg
becomes:
#main {
width: 70%; }
#main .note {
background-color: #55aaff; }
#main p {
width: 5em;
background-color: #55aaff; }
You can even do simple arithmetic operations with variables, adding numbers and even colors together:
!main_bg= #46ar12
!main_width= 40em
#main
:background-color= !main_bg
:width= !main_width
.sidebar
:background-color= !main_bg + #333333
:width= !main_width - 25em
becomes:
#main {
background-color: #46a312;
width: 40em; }
#main .sidebar {
background-color: #79d645;
width: 15em; }
Taking the idea of variables a bit further are mixins. These let you group whole swathes of CSS attributes into a single directive and then include those anywhere you want:
=blue-border
:border
:color blue
:width 2px
:style dotted
.comment
+blue-border
:padding 2px
:margin 10px 0
.reply
+blue-border
becomes:
.comment {
border-color: blue;
border-width: 2px;
border-style: dotted;
padding: 2px;
margin: 10px 0;
}
.reply {
border-color: blue;
border-width: 2px;
border-style: dotted;
}
A comprehensive list of features is in the documentation for the Sass module.
Indentation can be made up of one or more tabs or spaces. However, indentation must be consistent within a given document. Hard tabs and spaces can’t be mixed, and the same number of tabs or spaces must be used throughout.
The Haml gem includes several executables that are useful for dealing with Haml and Sass from the command line.
The haml executable transforms a source Haml file into HTML. See haml --help for further information and options.
The sass executable transforms a source Sass file into CSS. See sass --help for further information and options.
The html2haml executable attempts to transform HTML, optionally with ERB markup, into Haml code. Since HTML is so variable, this transformation is not always perfect; it’s a good idea to have a human check the output of this tool. See html2haml --help for further information and options.
The css2sass executable attempts to transform CSS into Sass code. This transformation attempts to use Sass nesting where possible. See css2sass --help for further information and options.
Haml and Sass are designed by Hampton Catlin (hcatlin) and he is the author of the original implementation. However, Hampton doesn’t even know his way around the code anymore and mostly just concentrates on the language issues. Hampton lives in Toronto, Ontario (though he’s an American by birth) and is a partner at Unspace Interactive.
Nathan Weizenbaum is the primary maintainer and architect of the “modern” Ruby implementation of Haml. His hard work has kept the project alive by endlessly answering forum posts, fixing bugs, refactoring, finding speed improvements, writing documentation, implementing new features, and getting Hampton coffee (a fitting task for a boy-genius). Nathan lives in Seattle, Washington and while not being a student at University of Washington he consults for Unspace Interactive and Microsoft.
Chris Eppstein is a core contributor to Sass and the creator of Compass, the first Sass-based framework. Chris focuses on making Sass more powerful, easy to use, and on ways to speed its adoption through the web development community. Chris lives in San Jose, CA with his wife and daughter. He is the Software Architect for Caring.com, a website devoted to the 34 Million caregivers whose parents are sick or elderly, that uses Haml and Sass.
If you use this software, you must pay Hampton a compliment. And buy Nathan some jelly beans. Maybe pet a kitten. Yeah. Pet that kitty.
Some of the work on Haml was supported by Unspace Interactive.
Beyond that, the implementation is licensed under the MIT License. Ok, fine, I guess that means compliments aren’t required.