What are some good ways to implement breadcrumbs on a Jekyll site?

I have improved slightly on the answers given earlier. I have removed the unordered list and seperated the items with a character (forward slash). I have added a filter for 'index.html' and '.html', so urls like 'mysite.com/path/index.html' and 'mysite.com/path/item-name.html' are also supported. Finally I have capitalized the titles. This results in something that looks like this:

Home / Path / Item name

{% assign crumbs = page.url | remove:'/index.html' | split: '/' %}

<a href="/">Home</a>
{% for crumb in crumbs offset: 1 %}
  {% if forloop.last %}
    / {{ crumb | replace:'-',' ' | remove:'.html' | capitalize }}
  {% else %}
    / <a href="{% assign crumb_limit = forloop.index | plus: 1 %}{% for crumb in crumbs limit: crumb_limit %}{{ crumb | append: '/' }}{% endfor %}">{{ crumb | replace:'-',' ' | remove:'.html' | capitalize }}</a>
  {% endif %}
{% endfor %}

PS. I have created an online resource for snippets like this: jekyllcodex.org/without-plugins

This should give breadcrumbs at any depth (with a caveat, see end). Unfortunately, the Liquid filters are fairly limited, so this is an unstable solution: any time /index.html appears, it is deleted, which will break URLs that have a folder that starts with index.html (e.g. /a/index.html/b/c.html), hopefully that won't happen.

{% capture url_parts %} {{ page.url | remove: "/index.html" | replace:'/'," " }}{% endcapture %}
{% capture num_parts %}{{ url_parts | number_of_words | minus: 1 }}{% endcapture %}
{% assign previous="" %}
<ol>
 {% if num_parts == "0" or num_parts == "-1" %}
  <li><a href="/">home</a> &nbsp; </li>
 {% else %}
  <li><a href="/">home</a> &#187; </li>

  {% for unused in page.content limit:num_parts %}
   {% capture first_word %}{{ url_parts | truncatewords:1 | remove:"..."}}{% endcapture %}
   {% capture previous %}{{ previous }}/{{ first_word }}{% endcapture %}

   <li><a href="{{previous}}">{{ first_word }}</a> &#187; </li>

   {% capture url_parts %}{{ url_parts | remove_first:first_word }}{% endcapture %}
  {% endfor %}
 {% endif %}
</ol>

How it works is:

• separates the URL, ignoring index.html (e.g. /a/b/index.html becomes a b, /a/b/c.html becomes a b c.html),
• successively takes and removes the first word of url_parts, to iterate through all but the last word (e.g. it goes a b c.html -> (a, b c.html) -> (b, c.html); then we stop).
• at each step, it makes the breadcrumb link using the current first_word, and previous which is all the previous directories seen (for the example above, it would go "" -> "/a" -> "/a/b")

NB. the page.content in the for loop is just to give something to iterate over, the magic is done by the limit:num_parts. However, this means that if page.content has fewer paragraphs than num_parts not all breadcrumbs will appear, if this is likely one might define a site variable in _config.yml like breadcrumb_list: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15] and use site.breadcrumb_list as the placeholder instead of page.content.

Here is an example (it doesn't use precisely the same code as above, but it's just a few little modifications).

I made the following breadcrumbs with liquid templates, its as elegant as I could manage :) It uses jQuery to set the active class on the li last li item.

            <ul class="breadcrumb">
                <li><a href="#"><i class="icon-home"></i>Home</a> </li>
                {% assign crumbs = page.url | split: '/' %}
                {% assign crumbstotal = crumbs | size %}
                {% for crumb in crumbs offset:2 %}
                    {% unless crumb == 'index.html' %}
                        <li><span class="divider">&#187;</span> {{ crumb | remove: '.html' }} </li>
                    {% endunless %}
                {% endfor %}
            </ul>
                            <script>$(".breadcrumb li").last().addClass('active');</script>