🦋 AJAX geekery
Permit me to wax geeky for a moment: last night I added a new feature to the site, which involves dynamic loading of page elements. Fun! About half (a little less) of the size of this page (ie, roughly 20K bytes) is the blogroll, on the right-hand side of the page under the "Where to go from here" heading -- which you might not think to look at it, since most of the data is hidden when the page loads and only shown when you click on the category headings. This means that whenever a person or a robot downloads the page, 20K of data is sent that is probably not going to be displayed or used. I've been trying for a while to figure out how to only send it to people who are interested in it, i.e. not to robots or to one-time visitors who come here from a Google search for a book their class is reading, which together account for the great majority of page views. Last night I came up with a pretty seamless solution: I recently implemented a sticky blogroll, using cookies to ensure that once you have clicked a blogroll category header, the category will remain visible when you reload the page. So it's easy to check whether a user is a repeat visitor who has in the past looked at the blogroll -- since the blogroll is stored in a file on the server that gets read at load time, all I had to do was create a truncated version of that file (containing only data for the default visible categories) and include that instead if the relevant cookie was not set. But then when a new visitor decides s/he wants to look at the blogroll, and clicks on a category header, how is s/he going to get data? Well in the truncated blogroll file, the empty headers are linked to a Javascript function called load_full_links, which uses XMLHttpRequest to download the complete blogroll -- so the first time you click a category you will see about a 1-second delay. But then your cookie is set, so going forward the blogroll is loaded with the page. I think it's a pretty nice bit of design/programming and I'm looking forward to using a similar algorithm for other pieces of functionality.
 Potential problems: - I have tested this with Firefox and Internet Explorer. I'm not sure how it will work with other browsers, and it will definitely not work in a browser where Javascript is disabled. This doesn't seem like a big deal to me but if it causes trouble for you, let me know and I'll try and come up with a solution.
- Search engines will no longer index most of my blogroll. This is, on one hand, good -- I have seen referrals from searches that hit an item on the blogroll, which generally seem like my page is not what the searcher was looking for -- and on the other hand, possibly not ideal -- Google and Technorati both pay a lot of attention to outgoing links.
posted morning of Saturday, March 29th, 2008
➳ More posts about The site
➳ More posts about Programming Projects
➳ More posts about Projects
|
