Sticking with my plans to avoid writing server-side code for this blog, I decided to write an offline search engine.
Yes, I realise it is silly, but I hadn’t done much AJAX and it looked fun. The publishing process generates a bunch of JSON search indexes in seperate files, the correct index gets loaded by the browser using XmlHttpRequest based on the search terms, and the Javascript generates the search results page.
[The dodgy IE3-esque code is for the benefit of Pocket IE on my Smartphone, which supports XmlHttpRequest, but not the DOM...]